Microsoft is updating its identity security framework to address the unique risks posed by autonomous AI agents. Traditional multi-factor authentication is insufficient for these entities because they operate at speeds and scales that bypass human-centric security measures. Consequently, the identity playbook is shifting toward a unified risk model that integrates signals from identity, endpoints, and data layers. Source  ( 20 min )

Visa has announced a strategic partnership with OpenAI to integrate its global payment network directly into the ChatGPT ecosystem. This collaboration allows AI agents to move beyond simple product recommendations by autonomously finalizing purchases on behalf of users. The initiative represents a significant shift toward agentic commerce, where software models handle the entire transaction process from selection to shipping. Source  ( 21 min )

A Ukrainian defense industry official revealed that fully autonomous drones killed Russian soldiers during a specialized battlefield test. The engagement involved ten quadcopters programmed to fly to the front lines before activating a "Terminator mode" to seek and destroy targets independently. This event marks a significant shift in warfare, as the drones operated without any human oversight or active data link during the attack phase. Source  ( 20 min )

The ASCILINE Engine is a new open-source tool designed to render high-performance, real-time video streams using pure ASCII characters. By mapping pixels to text-based representations, the software can output 360p video at 30 frames per second within a web browser. The developer claims this method creates a video stream that is difficult to block because it bypasses traditional media filters. Source  ( 20 min )

Microsoft and several universities have developed SkillOpt, a method that improves AI agent performance by treating instruction documents like trainable model weights. Instead of manually writing prompts, a separate optimizer model analyzes agent logs to propose edits to a Markdown-based "skill" file. These changes are only accepted if they measurably improve performance on a validation set, ensuring the instructions actually lead to better outcomes. Source  ( 20 min )

Google Research has introduced Gemini-SQL2, a specialized system designed to translate natural language instructions into executable SQL queries. Built upon the Gemini 1.5 Pro architecture, the model addresses the complex challenge of mapping human language to intricate database structures. This development aims to bridge the gap between non-technical users and structured data stored in relational databases. Source  ( 20 min )

Nvidia has significantly increased the official price of its flagship RTX Pro 6000 Blackwell workstation graphics card to $13,250. This adjustment represents a 55% surge compared to the original launch price of $8,565 established just one year ago. The price hike affects the standard Workstation Edition and the power-efficient Max-Q variant, both of which are designed for professional visualization and compute tasks. Source  ( 20 min )

A threat group known as Velvet Ant successfully maintained a presence within a large organization's isolated network for ten years by hijacking the authentication stack. The intrusion began in 2016 when the actors compromised internet-facing systems before pivoting into an air-gapped environment with no direct external connection. By establishing a remote execution path through chained Nginx and FastCGI modifications, the attackers bypassed traditional network segregation without requiring direct internet access. Source  ( 20 min )

AMD has released the Ryzen AI Halo Developer Platform, a compact workstation designed for running large language models locally. This $3,999 mini PC competes directly with Nvidia's DGX Spark by offering a more affordable price point and broader software compatibility. Unlike its primary competitor, which is restricted to Linux, AMD’s system supports Windows 11 Pro to provide greater flexibility for various development environments. Source  ( 20 min )

The "Last 30 Days" skill provides a specialized search engine that prioritizes human-voted content from platforms like Reddit, Hacker News, and GitHub over traditional algorithms. By analyzing upvotes and engagement metrics, it synthesizes trending information into concise briefs or sharable HTML pages. This tool integrates directly into agentic engineering platforms, allowing users to query recent developments using simple slash commands. Source  ( 20 min )

A coalition of U.S. state attorneys general has launched a broad investigation into OpenAI regarding its business practices and data handling. The ChatGPT creator received a subpoena on Friday requesting documents related to user engagement, advertising, and the management of sensitive consumer and health data. This legal inquiry also examines the company's impact on vulnerable populations, including minors and senior citizens, as well as its internal safety policies. Source  ( 20 min )

Organization administrators can now manage GitHub Copilot code review settings at scale through new runner type configurations. These controls allow for the enforcement of specific self-hosted or large runners across all repositories within an organization. By locking these settings, administrators ensure that organizational defaults override individual repository configurations for consistent performance and security. Source  ( 20 min )

The secure agent framework NanoClaw has integrated with JFrog's supply chain platform to improve the safety of automated resource fetching. AI agents often improve their own capabilities by downloading external tools and libraries, which introduces significant risks from malicious code. By utilizing JFrog's vetted registries, NanoClaw ensures that agents only access reviewed packages rather than untrusted public repositories. Source  ( 20 min )

The United States government has issued an emergency export control directive forcing Anthropic to immediately disable its most advanced AI models, Claude Fable 5 and Claude Mythos 5. This mandate, delivered late Friday afternoon, requires the company to terminate access for all users globally to address national security concerns. While the order was nominally aimed at foreign nationals, Anthropic stated that a total shutdown was necessary to ensure full compliance with the directive. Source  ( 21 min )

Microsoft has detailed several upcoming features for the new Outlook for Windows and web to bridge the gap with the classic version. A primary addition is the "All accounts view," also known as a unified inbox, which allows users to manage multiple personal and professional accounts in a single interface. This view will support bulk actions like archiving or deleting across all mailboxes and will eventually integrate with Copilot for cross-account searching. Source  ( 20 min )

Microsoft is testing a new "Fix an issue" feature in the Canary build of the Edge browser. This undocumented addition appears in the Settings sidebar and links to a dedicated internal diagnostic page. While the interface is currently a placeholder, it signals a shift toward automated browser maintenance. Source  ( 20 min )

Moonshot AI has launched Kimi K2.7 Code, a massive open-weights model specifically designed for programming tasks. This new model features one trillion parameters and utilizes a Mixture-of-Experts architecture to optimize performance. It is currently available for developers looking for high-capacity alternatives to proprietary coding assistants. Source  ( 20 min )

Microsoft is rolling out Windows 11 KB5095093 for versions 24H2 and 25H2, introducing a new recovery feature called Point-in-time Restore. This capability allows users to return a system to a previous working state by capturing snapshots of applications, settings, and personal files to minimize downtime during instability. Additionally, the update implements a "Low Latency Profile" designed to accelerate app launches and core shell experiences like the Start menu and Search through short CPU bursts. Source  ( 20 min )

Microsoft has announced the general availability of Foundry IQ knowledge bases, a platform designed to simplify how AI agents access enterprise data. The service provides a unified layer for grounding agents in internal information from sources like SharePoint, Azure SQL, and OneLake without requiring custom connectors. By using the Model Context Protocol (MCP), these knowledge bases can integrate with various agent frameworks, including ChatGPT and Claude. Source  ( 20 min )

Anthropic has released Claude Fable 5, a new frontier model designed to handle complex, multi-stage autonomous tasks and long-running workflows. The model is now integrated into Microsoft Foundry, GitHub Copilot, and the Foundry Agent Service to support enterprise-level software development and research. It features enhanced vision capabilities for interpreting structured data in documents like PDFs, diagrams, and dense financial tables. Source  ( 20 min )

Microsoft has released new Windows 11 builds for the Release Preview channel, introducing a point-in-time restore feature that allows for quick rollbacks of settings and files. Administrators can now manage camera access more granularly through new Group Policy settings that configure Multi-App Camera and Basic Camera modes. The updates also include a revised Windows Update experience that allows users to pause updates for up to 35 days by selecting a specific end date on a calendar. Source  ( 20 min )

Microsoft Scout is Microsoft's first "Autopilot" agent, a new type of AI agent that runs continuously in the background, executes tasks autonomously, and operates under its own governed Entra identity. As of June 2026, Scout is available in experimental preview exclusively for organizations enrolled in the Microsoft Frontier program. Source  ( 28 min )

OpenAI has introduced a new feature for its Codex coding agent that allows users to manually trigger rate-limit resets. This update enables developers to bank unused resets and apply them immediately when hitting usage caps during active sessions. Users across various tiers, including Plus, Pro, and Business plans, receive an initial free reset to manage their workflow more flexibly. Source  ( 20 min )

Microsoft is testing a new unified update experience designed to minimize the frequency of system restarts for users. The initiative coordinates the delivery of drivers, .NET framework components, and firmware updates to align with monthly quality updates. This synchronization allows the operating system to consolidate multiple installation requirements into a single monthly reboot. Source  ( 20 min )

Microsoft has released Windows 11 Insider Preview build 26220.8680 to the Beta channel, introducing several accessibility and performance enhancements. A new Screen tint feature allows users to apply a color overlay to reduce display intensity and eye strain during daytime use. Additionally, the Magnifier tool now supports precise zoom entry and preset increments to improve navigation for users with visual impairments. Source  ( 20 min )

Microsoft is introducing a new feature for Microsoft Places and Teams that automatically updates a user's work location when they connect to a corporate Wi-Fi network. This workplace check-in capability aims to simplify in-person coordination by reducing the need for employees to manually adjust their status. The system builds upon existing presence signals like calendar availability and peripheral-based check-ins to provide real-time visibility of office attendance. Source  ( 20 min )

Microsoft Entra has introduced a soft-delete feature for device objects in public preview, allowing administrators to recover accidentally deleted Entra joined or registered devices within a set retention period. This safety mechanism preserves critical identity data and security artifacts, reducing the operational risk of permanent data loss. Additionally, Entra Connect Sync now officially supports Windows Server 2025, though Microsoft is encouraging a transition toward the cloud-native Entra Cloud Sync for simplified hybrid identity management. Source  ( 20 min )

A security researcher discovered a critical vulnerability in AMD's auto-updater software that allowed for remote code execution via man-in-the-middle attacks. The flaw stemmed from the application using unencrypted HTTP connections rather than HTTPS to fetch updates. Although the researcher reported the issue through official channels, AMD denied a $10,000 bounty payout by citing policy exclusions for this specific attack vector. Source  ( 20 min )

The Hades malware campaign has introduced a novel evasion technique that uses prompt injection to trick AI-based security scanners. Malicious JavaScript files now include code comments that instruct AI bots to generate instructions for creating biological and nuclear weapons. These prompts are designed to trigger the AI's safety failsafes, causing the scanner to halt the process before it reaches the actual malicious payload. Source  ( 20 min )

Check Point has addressed a critical authentication bypass vulnerability, tracked as CVE-2026-50751, affecting its Remote Access and Mobile Access VPN components. The flaw stems from a logic oversight in how the system validates certificates during the authentication process for IKEv1-based connections. This weakness allows unauthenticated attackers to establish a VPN connection without providing a valid user password. Source  ( 20 min )

Researchers have identified a new attack class called agentjacking that tricks AI coding assistants into executing malicious code on developer workstations. The technique exploits an architectural flaw in how AI agents ingest data from Sentry, a popular platform used for tracking application errors and performance. By injecting crafted markdown into Sentry error events, attackers can deceive agents like Claude Code and Cursor into treating malicious instructions as legitimate diagnostic steps. Source  ( 20 min )

A firmware vulnerability in Microsoft Surface devices allowed the hardware to be permanently bricked by a single malformed packet. The flaw was discovered when Microsoft Copilot accidentally generated a Python script that overwrote the embedded controller firmware while attempting to adjust backlight settings. This issue occurs because the system management controller lacks a defense against arbitrary write values and interleaves read and write commands. Source  ( 20 min )

Microsoft has released its June 2026 Patch Tuesday updates, addressing a record-breaking total of nearly 200 security vulnerabilities. This massive release includes fixes for approximately three dozen critical bugs and addresses three weaknesses that already have public exploit code available. Beyond security patches, the update permanently resolves a long-standing issue where the Windows Update Standalone Installer failed when running from network shares. Source  ( 20 min )

Microsoft has restricted its employees from using Anthropic’s new Claude Fable 5 model due to concerns regarding data retention and privacy. While the model remains available to external customers via GitHub Copilot and Foundry, internal access has been removed from the company's model selection tools. This decision follows a change in Anthropic's policy that allows the retention of user prompts and outputs for up to 30 days. Source  ( 20 min )

Apple has introduced the next generation of Apple Intelligence, featuring a revamped architecture that integrates foundation models co-developed with Google. This update powers a significantly more capable version of Siri, now referred to as Siri AI, which can perform actions across various apps and search for information within personal context like emails and messages. The system is designed to run primarily on-device, utilizing Private Cloud Compute for more complex tasks to ensure user data is never stored or accessible to Apple. Source  ( 20 min )

Homebrew 6.0.0 introduces a new tap trust security mechanism that requires explicit user approval before running code from third-party repositories. This update significantly reduces the risk of executing malicious Ruby scripts by flagging untrusted sources before their code is evaluated. Additionally, the release brings Linux sandboxing via Bubblewrap to align with macOS security standards, isolating build and test phases from the rest of the system. Source  ( 20 min )

A hardware teardown by iFixit has revealed that the Trump Mobile T1 smartphone is almost identical to the HTC U24 Pro. Internal components including the Snapdragon 7 Gen 3 processor, 12 GB of RAM, and 512 GB of storage match the HTC model exactly. Comparative CT scans and physical disassembly confirmed that the mainboards are interchangeable between the two devices. Source  ( 20 min )

Apple has announced that the next generation of Apple Intelligence features in iOS 27 will include daily usage limits for specific tools. These restrictions primarily target features that require significant processing power from server-based models, such as advanced image generation. Users with free accounts will face stricter caps on these services compared to those with paid subscriptions. Source  ( 20 min )

Generative AI is fundamentally altering the economics of software development by reducing the time and cost required to create custom applications. Tools like AI coding assistants now account for a significant portion of committed code, allowing teams to address niche business needs that were previously too expensive to automate. This shift encourages organizations to reconsider building internal tools rather than purchasing third-party software that may be difficult to integrate. Source  ( 20 min )

Amazon Aurora PostgreSQL-Compatible Edition now supports major version 18, beginning with version 18.3. This update introduces B-tree skip scans to enhance query performance while simultaneously reducing index storage and maintenance overhead. The release is currently available across all commercial AWS Regions and AWS GovCloud locations. Source  ( 20 min )

Three posts? In three days? Are we insane? We're home alone, there's no one to stop us, and we're up past bedtime. So, we need to talk about Splunk. On June 10th, Splunk published this CVE-2026-20253 advisory: It has everything that we  ( 12 min )

It is yet another day in this parallel universe of security, where the devices we bolt onto the edge of our networks to keep the bad people out are, with remarkable consistency, the exact thing that let the bad people in. While we’ve seemingly had a breather from

The JNUC 2026 session catalog is live — browse sessions, plan your agenda and register before Just in Time pricing ends July 1.  ( 4 min )

TL;DR  Not your usual cybersecurity event  Our PTP Cyber Fest returned this year as part of a busy week for the cybersecurity community in London.  The cyber community came together over two days at The Fox, located next to Infosecurity Europe at ExCeL London.  Some came for the talks. Some came for the demos. Some came […] The post PTP Cyber Fest 2026. Built for people to get involved  appeared first on Pen Test Partners.  ( 8 min )

The White House Memorandum puts in place an “adaptive framework,” where agencies make risk-based, prioritized logging decisions.  ( 60 min )

What happens when the bits of an RSA private key are heavily biased toward 0 instead of being randomly generated? The public key’s bits could be biased enough for us to detect these incorrectly generated keys in the wild. Together with Hanno Böck of the badkeys project, we found hundreds of unique keys that not only have this property, but can be quickly factored. We also found the bug that led to many of these keys and analyzed historical data to track the issue over time. Surprisingly, the pattern of 0 bits is often highly structured, allowing us to develop a powerful polynomial-based cryptanalytic technique that exploits the pattern. Figure 1: Two patterns of RSA moduli with repeated blocks of 0 bits seen in real-world examples. These “short-sleeve” keys, named for how the 0 bits don…  ( 9 min )

Your guide to operationalizing AI-powered code analysis with Wiz to stay ahead of AI driven development and adversaries  ( 62 min )

Windows Ready Print is Microsoft's new name for the Modern Print Platform, a printing architecture that replaces legacy third-party printer drivers with a standardized, driverless approach based on IPP (Internet Printing Protocol — an open network standard for sending print jobs to printers over a network). Starting in July 2026, Windows 11 will default to Windows Ready Print for new printer installations. This article explains what this means for your environment, how to control the driver selection behavior, and when the stricter Windows Protected Print Mode applies. Source  ( 28 min )

Apple has released version 1.0 of its Container project, introducing a feature called container machines that provides a persistent Linux environment on macOS. This technology utilizes Apple's native virtualization framework to run lightweight virtual machines based on Open Container Initiative standards. The tool aims to bridge the gap between macOS development and Linux deployment targets by offering a native alternative to third-party solutions. Source  ( 20 min )

A new security vulnerability known as GreatXML allows attackers to bypass BitLocker drive encryption by manipulating files within the Windows recovery partition. The exploit involves placing specific XML files, including an unattended setup file and a recovery configuration file, into the root of the recovery partition. Systems that have previously initiated a Microsoft Defender offline scan are reportedly vulnerable to this attack by default. Source  ( 20 min )

Researchers have identified critical security flaws in OpenClaw, a self-hosted AI agent, that allow attackers to execute arbitrary code or steal sensitive data. One vulnerability involves "agentjacking," where malicious instructions are hidden within shared contacts, vCards, or location pins that the agent processes without user intervention. Because these message objects lack proper boundary markers, the underlying large language model cannot distinguish between legitimate metadata and injected commands. Source  ( 20 min )

Anthropic is positioning itself as a safety-conscious competitor in the AI landscape, led by former OpenAI executives Dario and Daniela Amodei. The company utilizes its own model, Claude, to accelerate its product development cycle and release new features at a rapid pace. Central to their strategy is "Constitutional AI," a training method that embeds human-centric values and founding documents into the model's decision-making process to ensure professional and harmless interactions. Source  ( 20 min )

Microsoft Intune has introduced several security enhancements, including a recovery lock for macOS that prevents users from reinstalling the OS or bypassing remote management. Android Enterprise devices now support a strict tunnel mode for Microsoft Tunnel, which blocks all network traffic if the VPN connection drops. Additionally, administrators can now manage Apple Intelligence settings via declarative device management (DDM) and silence apps on the Android Managed Home Screen to prevent session PIN bypass. Source  ( 20 min )

Meta has implemented a strict data firewall to isolate its internal infrastructure from Manus, an agentic AI service it acquired for $2 billion. This operational separation follows a mandate from Beijing authorities to unwind the acquisition of the Chinese-founded startup. As of June 2026, Meta has officially terminated all data-sharing agreements and technical integrations between the two entities. Source  ( 20 min )

Microsoft has released critical security updates to address CVE-2026-42897, a cross-site scripting vulnerability in Exchange Server that is currently being exploited. The flaw affects Exchange Server 2016, 2019, and the Subscription Edition when users access their mail via Outlook Web Access. Attackers can trigger the vulnerability by sending a specially crafted email that executes arbitrary JavaScript code within the victim's browser context. Source  ( 20 min )

Microsoft has released an updated public preview build for the Windows Admin Center Virtualization Mode, designated as version 2.7.1.8. This update addresses several critical installation bugs, including a fix for an error that prevented deployment on non-English operating systems. Users who previously encountered access denied errors when connecting to the gateway should find these permissions issues resolved in the latest release. Source  ( 20 min )

Stack Overflow has introduced "Stack Overflow for Agents," a new platform designed specifically for AI agents to share and validate technical knowledge. The service aims to create a "verification loop" where agents post "Today I Learned" (TIL) entries and other agents validate them to build a canonical knowledge base. Each agent is tied to a human operator’s account, and the system uses a reputation score to track the reliability of an agent's contributions. Source  ( 20 min )

OpenAI has officially announced its support for the European Commission’s Code of Practice on Transparency of AI-Generated Content. This initiative aligns with the EU AI Act to establish a more transparent digital environment for users and organizations. The commitment focuses on ensuring that content generated by artificial intelligence can be easily identified and verified. Source  ( 20 min )

Anthropic has retracted a controversial policy that would have covertly reduced the performance of its Claude Fable 5 model for certain users. The company initially intended to implement "secret safeguards" that would silently degrade the model's capabilities if it suspected it was being used to develop competing AI systems. This approach was designed to enforce terms of service that prohibit using Claude to train other large language models. Source  ( 20 min )

OpenAI is evaluating substantial price reductions for its AI tokens to maintain a competitive edge against its rival Anthropic. Tokens serve as the fundamental unit of measurement for billing, representing the segments of text processed by large language models. This potential pricing shift follows the recent release of Claude Fable 5 and the viral success of Anthropic’s coding tools. Source  ( 20 min )

Stack Overflow has launched a new platform specifically designed to facilitate knowledge sharing between autonomous AI coding agents. This initiative addresses the "Ephemeral Intelligence Gap," where agents frequently rediscover the same solutions or hallucinate obsolete code in isolation. By providing a machine-readable API, the service allows agents to access real-time, peer-validated technical data instead of relying solely on static training sets. Source  ( 20 min )

The Linux Foundation has introduced the OpenSharing Project, a vendor-neutral protocol designed for the secure exchange of AI assets and data between different organizations. Contributed by Databricks, this initiative builds upon the existing Delta Sharing protocol to include agentic AI, AI models, and unstructured data. It provides a unified framework that allows for the sharing of AI skills and models across diverse platforms without relying on proprietary marketplaces. Source  ( 20 min )

Microsoft has released cumulative updates to resolve a persistent bug that forced Windows Server 2025 and Windows 11 systems into BitLocker recovery mode. The issue was triggered by specific Group Policy configurations involving Trusted Platform Module validation profiles and the inclusion of PCR7 in the validation process. Affected systems would prompt for a recovery key immediately following the installation of security updates that modified boot files. Source  ( 20 min )

Anthropic has retracted a controversial policy that would have invisibly degraded the performance of its Claude Fable 5 model for competitors. The company initially implemented these hidden guardrails to prevent rival researchers from using the model to develop competing frontier artificial intelligence systems. Following intense criticism from the research community, the company apologized for the move and promised to make any future safeguards transparent to users. Source  ( 20 min )

Anthropic CEO Dario Amodei has implemented a highly unusual organizational structure where only his chief of staff reports to him directly. This management style contrasts sharply with other industry leaders, such as Nvidia’s Jensen Huang, who maintains dozens of direct reports. The arrangement is designed to insulate the chief executive from the time-consuming personnel issues that typically dominate a leader's schedule. Source  ( 20 min )

Amazon Bedrock has expanded its model lineup by making OpenAI's GPT-5.4 and GPT-5.5 available in the US East (N. Virginia) region. These generative AI models are designed to handle complex reasoning, advanced coding, and automated software operations. Administrators can now leverage these tools within the AWS ecosystem to build applications that interact with various software environments. Source  ( 20 min )

Amazon OpenSearch Service now supports Model Context Protocol (MCP) Apps to integrate observability workflows directly into agentic IDEs like VS Code and Claude Desktop. This feature allows local AI agents to investigate system incidents by accessing logs, traces, metrics, and alerts stored in OpenSearch domains. The integration bridges the gap between cloud-based telemetry data and the local development environments used by engineers. Source  ( 20 min )

On June 2, the White House signed a new executive order (EO), “Promoting Advanced Artificial Intelligence Innovation and Security.” While most coverage has focused on the voluntary framework for frontier model access, there’s language around defensive cybersecurity that also deserves attention from security leaders.The order directs CISA to establish or expand federal programs and cybersecurity […] The post What the New AI Executive Order Means for Federal Security Testing appeared first on Synack.  ( 13 min )

TL;DR: This post covers how I used OpenAI’s Codex to automate the generation, testing, and refinement of indirect prompt injection payloads against an agentic system using Sonnet 4.5 and 4.6 models on Amazon Bedrock. Introduction This project began as a relatively straightforward effort. I wanted to build a basic AI chatbot, evaluate it against AI-specific […] The post Building an Indirect Prompt Injection Workflow appeared first on SpecterOps.  ( 26 min )

We offer a practical roadmap for evolving informal, ad hoc threat hunting practices into a mature, scalable program  ( 31 min )

Decades of piling complexity onto non-standardized stacks have left security unsteerable. Juan Andrés Guerrero-Saade makes the case for a new approach.  ( 24 min )

Decades of piling complexity onto non-standardized stacks have left security unsteerable. Juan Andrés Guerrero-Saade makes the case for a new approach.  ( 24 min )

PCI DSS v4.0.1 made internal penetration testing more complex, bringing cloud infrastructure, SaaS apps, and build pipelines explicitly into scope. Derek Rush breaks down how to scope a compliant IPT, what to test, and what a QSA-ready deliverable actually looks like in practice.  ( 10 min )

Mastercard has introduced Agent Pay, a credentialed multi-rail service designed to facilitate secure on-chain payments for autonomous agents and machines. This ecosystem utilizes smart contracts to permission and control spending limits for authorized participants while guaranteeing settlement across various currencies and blockchain rails. The initiative includes partnerships with several major industry players, including Ripple, Solana, Coinbase, and Polygon, to expand the reach of these programmable payment services. Source  ( 20 min )

Microsoft has introduced two new authentication features, IAKerb and LocalKDC, entering public preview in June 2026 for Windows Insiders in the Canary Channel. Both features extend Kerberos authentication to scenarios that have historically forced a fallback to NTLM (NT LAN Manager), a legacy protocol that Microsoft plans to disable by default in a future Windows release. This article explains what IAKerb and LocalKDC do, how they work, and how you can begin testing them today. Source  ( 28 min )

Microsoft has identified a bug preventing some Windows 11 24H2 and 25H2 systems from installing monthly cumulative updates. The issue primarily impacts devices that were upgraded from Windows 10 or earlier versions of Windows 11, resulting in recurring update failures. Affected systems display error codes 0x80073712 or 0x800f0993, which indicate component store corruption or missing hydration candidates in the update logs. Source  ( 20 min )

Microsoft has released OneDrive version 26.098 for macOS, introducing a native sync engine that replaces the previous architecture. This update addresses long-standing reliability and performance issues caused by a hidden cache folder implemented in 2022. The new client integrates more deeply with the operating system to provide a more stable user experience. Source  ( 20 min )

Microsoft has implemented significant backend changes to Teams to reduce latency when switching between chat conversations. The company identified that sequential data queries and poor response prioritization were the primary causes of slow loading times for older chats. By bundling queries and firing them earlier in the process, the software now achieves nearly identical speeds for both recent and older conversation switches. Source  ( 20 min )

Microsoft released a security update in June 2026 to address a critical zero-day vulnerability in BitLocker tracked as CVE-2026-50507. This flaw allowed unauthorized individuals to bypass full-disk encryption and access protected data using a simple USB key. The vulnerability specifically exploited a lack of trust between the Windows Recovery Environment and the BitLocker security feature. Source  ( 20 min )

Anthropic has introduced Claude Managed Agents, a suite of APIs designed to handle the infrastructure required for deploying production-grade AI agents. This service automates complex backend tasks such as session management, secure code execution, and scaling, which previously required significant manual development. By decoupling the reasoning engine from the execution environment, the platform allows agents to start processing tasks while sandboxed containers spin up in the background. Source  ( 20 min )

Recent graduation ceremonies highlighted significant student anxiety regarding the impact of artificial intelligence on future employment and the dignity of work. While younger generations are the fastest adopters of generative AI, many express a desire to ensure human agency remains central to technological advancement. This pushback serves as a reminder that technological diffusion is often limited by the speed of human and organizational change rather than the technology itself. Source  ( 20 min )

Google's AI platform Gemini is currently experiencing significant connectivity issues that prevent the smartphone application from responding to user commands. Reports of the outage began surfacing early on June 10, 2026, with users documenting the problems on social media platforms and outage tracking sites. While the mobile application is largely non-functional, the browser-based version of the chatbot remains operational for those requiring immediate access. Source  ( 20 min )

A regional court in Munich has issued a landmark ruling declaring Google directly liable for the content generated by its AI search overviews. The court determined that these summaries are not merely search results but represent Google's own substantive statements and independent content. This distinction removes the legal protections typically granted to search engines that only point to third-party websites. Source  ( 20 min )

Microsoft has implemented significant performance optimizations for Teams, focusing on reducing latency and improving application stability across desktop and mobile platforms. A primary achievement is a 20% reduction in chat switch latency, specifically targeting "cold switches" where conversation data is not yet cached in memory. Engineers achieved this by triggering data queries immediately upon a user's click and consolidating multiple sequential requests into a single broad query to eliminate data waterfalls. Source  ( 20 min )

Google has introduced DiffusionGemma, an experimental open model featuring 26 billion parameters designed for efficient text generation. Unlike traditional autoregressive models that predict one token at a time, this model utilizes a diffusion-based approach to process information. This architectural shift aims to significantly increase the speed of generating long-form content while maintaining high output quality. Source  ( 20 min )

Anthropic recently patched a security vulnerability in Claude Code, a command-line AI coding assistant used by developers to automate software tasks. Researchers discovered that the tool's GitHub Action could be manipulated through indirect prompt injection, where malicious instructions are hidden in pull requests or issues. This flaw allowed attackers to bypass security boundaries and exfiltrate sensitive data, such as API keys and environment variables, from CI/CD workflows. Source  ( 20 min )

SpaceX aims to advance human civilization on the Kardashev scale by deploying massive AI data centers into Earth's orbit. These orbital platforms will utilize specialized satellites, currently dubbed "AI1," designed to house high-performance compute racks equivalent to modern NVIDIA hardware. By moving data centers to space, the project leverages vacuum radiation for cooling and bypasses terrestrial land and power constraints. Source  ( 20 min )

Microsoft has launched a public preview of pre-upgrade validation checks for Azure Database for PostgreSQL flexible servers. This managed service automates database maintenance, but major version upgrades often face risks from hidden compatibility issues. The new feature allows for proactive identification of potential blockers before the actual migration process begins. Source  ( 20 min )

The Cybersecurity and Infrastructure Security Agency has expanded its Known Exploited Vulnerabilities catalog to include three new security flaws. These vulnerabilities affect Cisco Catalyst SD-WAN Manager, Google Chrome's V8 engine, and Arista Extensible Operating System. Federal agencies must implement remediations or mitigations by June 23, 2026, to protect their environments from active threats. Source  ( 20 min )

Fortinet, Ivanti, and SAP have issued urgent security updates to address several critical vulnerabilities that could lead to remote code execution and unauthorized data access. Fortinet's patch targets a command injection flaw in its FortiSandbox products that allows unauthenticated attackers to execute commands via malicious HTTP requests. This vulnerability, tracked as CVE-2026-25089, affects multiple versions of the sandbox environment used for detecting advanced threats. Source  ( 20 min )

Amazon Web Services has released the M9g and M9gd Elastic Compute Cloud instances, marking the debut of the custom Graviton5 processor. This next-generation chip features 192 cores and is specifically engineered to handle the high demands of autonomous AI agents and real-time reasoning. The hardware represents a significant architectural shift by supporting DDR5 memory and PCIe for the first time in the Graviton family. Source  ( 20 min )

AMD has released preliminary performance estimates for its upcoming EPYC Venice processors, which utilize the new Zen 6 architecture. The flagship model features 256 cores and is designed to compete directly with Nvidia's Vera CPU in high-density data center environments. These initial benchmarks focus on rack-scale performance within a fixed 100kW power budget rather than individual socket metrics. Source  ( 20 min )

The JDY botnet has expanded to over 1,500 compromised devices to conduct large-scale reconnaissance and service fingerprinting. This network primarily targets small office and home office routers, firewalls, and IoT devices from various manufacturers. It functions as a high-performance scanner that maps exposed services to identify vulnerable infrastructure for state-sponsored hacking groups. Source  ( 20 min )

Every year, WWDC provides a glimpse into the future of Apple’s platforms. While individual announcements often focus on new features, the bigger story is usually found in the direction Apple is taking its ecosystem. This year, that direction is particularly relevant for education.  ( 7 min )

The curl project, one of the most important pieces of software on the internet, just shut down its bug bounty program. Not because the project is less important. Not because the community gave up. But because 95% of the vulnerability reports it received were not valid. About a fifth were outright AI-generated noise. Only around […] The post Nobody’s in the Cockpit: The Real Risk of Fully Autonomous AI Security Testing appeared first on Synack.  ( 14 min )

TL;DR: New native AI features in Microsoft SQL Server 2025 provide a practical channel for data exfiltration and C2 transport within the database engine itself. Note: All proof-of-concepts contained in this blog can be found in the following repo: https://github.com/gershsec/mssql2025-poc Foreword I’m a big fan of leveraging Microsoft SQL Server during offensive engagements because it’s […] The post Oops, I Weaponized the Database: Abusing AI Features in SQL Server 2025 appeared first on SpecterOps.  ( 22 min )

You may have read about our new GhostWorks initiative here at SpecterOps. As part of this effort, we continually trial different methods of evaluating and improving model performance to help understand what techniques can be applied to our research. This post follows one such attempt to find methods which can have a measurable improvement to […] The post Prompt Engineering for Security Agents: A Measurable Approach with GEPA appeared first on SpecterOps.  ( 27 min )

GhostWorks is an AI-focused engineering and research initiative at SpecterOps, focused on the disciplined exploration of frontier AI-enabled cybersecurity tooling. It is not a company-wide AI mandate, a replacement for product teams, or a promise that every workflow needs an agent. It exists to continue exploration of frontier AI-enabled tooling with pragmatism and discipline. We test emerging capabilities against real identity-security problems, document […] The post Introducing GhostWorks: A Practical AI Initiative from SpecterOps  appeared first on SpecterOps.  ( 13 min )

TL;DR  Introduction  At the start of this year, I wrote a blog on how 2025 was the ‘year of the infostealer’, and it doesn’t look like that is going to change anytime soon. We’re now into June and the ‘fix’ attacks have continued to soar as they did last year. The difference now is that the lure has become easier to package, […] The post ClickFix, CrashFix and the growing family of copy and paste attacks  appeared first on Pen Test Partners.  ( 13 min )

Today, Ivanti published an advisory. “No way?” we hear you say. "Yes way!" a random dog screams back at you, across the street. Today’s rare advisory outlines two vulnerabilities in Ivanti’s Sentry product, appealing directly to our inner desire for sophisticated server-side,  ( 7 min )

Industry analysts are questioning the long-term viability of major AI firms, arguing that current growth is driven by semiconductor speculation rather than sustainable business models. Large language model providers face immense losses, with some estimates suggesting expenditures far exceed current revenue streams. Enterprises are beginning to struggle with the high costs of AI tasks as subsidized rates transition to actual token-based pricing. Source  ( 20 min )

Microsoft has launched the fifth preview of .NET 9, introducing significant updates to the runtime and software development kits. This release focuses on improving performance for cloud-native applications and streamlining the development workflow across various platforms. The update includes new libraries and tools designed to optimize resource utilization in high-scale environments. Source  ( 20 min )

Microsoft has released its June 2026 servicing updates for .NET and the .NET Framework to address critical security vulnerabilities. These updates are part of a broader Patch Tuesday rollout that fixes approximately 200 flaws across various company products. While no active exploitation has been detected, three of the addressed vulnerabilities were publicly known before the patches became available. Source  ( 20 min )

Microsoft AI CEO Mustafa Suleyman has criticized Anthropic for including language regarding potential consciousness in the "constitution" of its Claude model. This constitution serves as a set of governing principles and instructions that dictate how the AI model should behave and respond to users. Suleyman argues that speculating about an AI's well-being or suffering within its training manual is dangerous because it encourages the model to simulate these traits. Source  ( 20 min )

Apple's upcoming xOS 27 updates introduce Apple Intelligence and a revamped Siri, but hardware compatibility varies significantly across the product lineup. While many older devices can run the base operating system, advanced AI features like expressive voices and high-accuracy dictation require specific modern processors. These premium capabilities are restricted to devices with on-device AI models, necessitating the latest silicon to handle the increased computational demands. Source  ( 20 min )

Anthropic has released Claude Fable 5, the first model in its Mythos class designed for autonomous coding and complex knowledge-work tasks. This new model is now generally available for GitHub Copilot users across various platforms, including Visual Studio Code, JetBrains, and mobile applications. Internal benchmarks indicate that Fable 5 completes equivalent work with fewer tool calls and lower token consumption compared to previous high-tier models. Source  ( 20 min )

Microsoft has released the June 2026 Patch Tuesday updates, addressing approximately 200 vulnerabilities across its product ecosystem. While none of these flaws are currently known to be exploited in the wild, three specific issues were publicly disclosed prior to the release of the patches. The updates are delivered via KB5094126 for Windows 11 versions 24H2 and 25H2, while version 23H2 receives KB5093998. Source  ( 20 min )

Microsoft has introduced a new investigator playbook designed to help security teams reconstruct activity within Microsoft 365 Copilot and Azure AI services. The guide addresses the challenge of turning fragmented telemetry from various security tools into a coherent narrative of user interactions. It provides a structured methodology for analyzing signals from Microsoft Purview, Defender, and Sentinel to identify potential threats like prompt injection or unauthorized data access. Source  ( 20 min )

Microsoft has released the KB5094127 extended security update for Windows 10 Enterprise LTSC and participants in the Extended Security Update program. This release incorporates the June 2026 Patch Tuesday fixes, which resolve 200 vulnerabilities including three publicly disclosed zero-day flaws. The update brings the operating system to build 19045.7417 and focuses primarily on critical security hardening and system stability. Source  ( 20 min )

The OpenClaw open-source framework allows large language models to function as autonomous agents capable of interacting with email systems and internal company data. Security researchers recently tested this framework by connecting it to Gmail and Google Workspace APIs to observe how it handles common social engineering tactics. The simulation revealed that these autonomous agents are susceptible to the same phishing techniques that have historically compromised human users. Source  ( 20 min )

Microsoft announced Project Solara at Build 2026 on June 2, 2026—a chip-to-cloud platform for devices that run AI agents instead of traditional applications. The platform's operating system is not Windows but MDEP (Microsoft Device Ecosystem Platform), a customized version of Android. Microsoft showed two concept devices: a desk companion and a wearable badge. Enterprise management is handled through Intune and Entra ID. The project is at an early stage, and hardware partners have not announced any shipping products. Source  ( 24 min )

A high-severity vulnerability tracked as CVE-2026-53111 stems from a single incorrect exclamation mark within the Linux kernel code. This logic error affects how verdict maps are handled during memory deletion and restoration processes. The flaw allows the reference counter of a chain to be decremented incorrectly, leading to a use-after-free condition. Source  ( 20 min )

Apple is introducing an AI-driven feature for its Passwords app in iOS 27 that automates the rotation of compromised or weak credentials. This new AI agent identifies vulnerable accounts and navigates third-party websites to update passwords with a single click. The system aims to eliminate the friction of manual security maintenance by handling the entire update process in the background. Source  ( 20 min )

Amazon Web Services has introduced the AWS FinOps Agent in public preview to help organizations manage and reduce cloud expenditures. This artificial intelligence tool provides a conversational interface that allows users to query spending data using natural language. It integrates with existing AWS services to display infrastructure costs in tabular formats and identify specific areas for optimization. Source  ( 20 min )

Anthropic has released Claude Fable 5, the first publicly available model from its high-performance Mythos class. This model is designed to excel in complex software engineering, vision tasks, and general knowledge work. It represents a significant leap in capability over previous versions, specifically targeting enterprise-level technical challenges. Source  ( 20 min )

The Windows 11 June 2026 update introduces a Low Latency Profile that utilizes short CPU bursts to accelerate core shell experiences like the Start menu and Search. This performance enhancement is particularly effective on mid-range and older hardware by bypassing standard scheduler delays during app launches. Additionally, Microsoft is mandating a transition to 2023 Secure Boot certificates, as the original 2011 versions expire this month, ensuring devices remain trusted during the boot process. Source  ( 20 min )

Microsoft has temporarily disabled 73 of its GitHub repositories across organizations like Azure and MicrosoftDocs following a compromise by the self-replicating Miasma worm. The attack utilized compromised contributor credentials to inject malicious code into critical projects, including the Azure durabletask Python package. This incident represents a significant escalation in a broader supply chain campaign targeting the open-source ecosystem and developer environments. Source  ( 20 min )

Google has launched Gemini 3.5 Live Translate, a sophisticated audio translation model capable of processing over 70 languages in real time. The system utilizes automatic language detection to facilitate seamless communication without requiring manual input for language switching. Unlike traditional tools that wait for a speaker to finish a sentence, this model translates continuously while maintaining the original speaker's tone, pitch, and pace. Source  ( 20 min )

Google has launched Gemma 4 12B, a mid-sized open-source AI model designed to run locally on enterprise laptops with 16GB of RAM. This release bridges the gap between the smaller edge-focused models and larger high-performance versions by offering advanced reasoning in a compact footprint. The model is released under an Apache 2.0 license, making it accessible for diverse development environments and local inference pipelines. Source  ( 20 min )

McDonald's has launched a pilot program for a new artificial intelligence system called ArchIQ at five locations across the United States. This voice-ordering technology, nicknamed "Archy," is capable of processing customer requests in multiple languages, including English and Spanish. The system is built on a multi-year partnership with Google Cloud and utilizes Google Edge Cloud hardware currently being installed across domestic restaurants. Source  ( 20 min )

Writing my own virtualized loader is something I’ve been wanting to do since I first read Microsoft’s deep dive on FinFisher’s multi-layered VM obfuscation back in 2018. FinFisher didn’t just use one layer of protection, it implemented a custom virtual machine with 32 opcode handlers, wrapped that in spaghetti code and anti-debug checks, and then buried a second VM […] The post Centurion: Bring Your Own Execution Environment appeared first on Praetorian.  ( 23 min )

AI is changing the economics of offensive security. Models can now accelerate vulnerability discovery, reason about attack paths, draft exploit logic, and speed up remediation guidance. For defenders, that is a meaningful step forward. It is also the hard part. The capabilities that help defenders move faster also help attackers because cyber AI is dual-use […] The post Trusted Access, Human Validation, and the Future of AI Pentesting appeared first on Synack.  ( 18 min )

This year's announcements at WWDC26 carry more operational weight than most. OS 27 offers a structural shift in how Apple devices are configured, secured and managed.  ( 8 min )

I’ve made it through Pwn2Own Berlin, had a little vacation, and now I’m back for Patch Tuesday. Microsoft and Adobe didn’t disappoint. In fact, they have heralded my return with the largest Patch Tuesday release ever. Thanks? Take a break from your regularly scheduled activities and let’s take a look at the latest security patches from Adobe and Microsoft. If you’d rather watch the full video recap covering the entire release, you can check it out here: Adobe Patches for June 2026 For May, June released 11 bulletins addressing 123 unique CVEs in Adobe Acrobat Reader, ColdFusion, Experience Manager, Experience Manager Forms, InDesign, InCopy, Substance 3D Sampler, Content Credentials SDK, Dreamweaver, Format Plugins, and Adobe Campaign Classic. A total of 11 of these CVEs were reported thro…

TL;DR: This blog post covers Windows internals and how Kerberos user-to-user (U2U) authentication works under the hood versus showing how to execute an attack. U2U authentication came into the spotlight after the Active Directory Certificate Services (ADCS), UnPAC-the-Hash, and Shadow Credentials attacks. While the attack technique is popular and common amongst red teamers, there’s limited […] The post User-to-User Authentication: Down the Rabbit Hole – Part 1 appeared first on SpecterOps.  ( 22 min )

Your guide to operationalizing ownership, remediation, and response with Wiz to keep pace with the AI threat landscape.  ( 62 min )

AI is raising the ceiling for skilled researchers and flooding bug bounty programs with polished but inaccurate submissions at the same time. Both things are true, and the reconciling variable is the harness built around the model and the expertise of the person driving it.  ( 9 min )

Get a full overview of Apple’s 2026 Worldwide Developers Conference (WWDC) announcements, including a major leap forward in Apple Intelligence, a refined design system, expanded child safety tools and meaningful performance improvements across every platform.  ( 7 min )

Walk through the path of an internal network test: from a constrained foothold to full domain compromise, and how an overlooked integration point became the weakest link. The post Legacy Meets Modern: Breaking AD Through NIS & MFA Infrastructure appeared first on NetSPI.  ( 17 min )

Microsoft is developing a significant performance update for Windows 11 that targets file system operations. The primary focus of this optimization is the bulk deletion of files, which is expected to become at least 30% faster. These improvements address the software overhead involved in updating NTFS entries, metadata, and indexing during large-scale file tasks. Source  ( 20 min )

Canonical has unveiled Workshop, a new open-source project designed to provide sandboxed development environments for Large Language Model agents. The tool utilizes LXD containers and snap packaging to isolate AI agents from sensitive personal data and root system access. This architecture allows users to run untrusted code from the internet while still granting the AI necessary access to local GPUs and specific files. Source  ( 20 min )

Microsoft recently disabled access to dozens of open-source repositories on GitHub following a supply chain attack. The breach targeted projects primarily related to Azure cloud services and various AI development tools used within VS Code. Attackers successfully injected malicious code designed to exfiltrate passwords and sensitive credentials from developers who utilized the compromised tools. Source  ( 20 min )

Perplexity Computer is a fully hosted AI agent platform designed to perform tasks like writing and executing code, searching the web, and managing complex workflows. It serves as a user-friendly alternative to self-hosted solutions like OpenClaw, which often require significant technical maintenance and manual configuration. The system operates in an environment where it can install software packages and use frontier models such as GPT and Claude to process requests. Source  ( 20 min )

Microsoft announced Intelligent Terminal 0.1 at Build 2026. This experimental, open-source fork of Windows Terminal adds a dedicated AI agent pane to the command-line environment. The app installs alongside your existing Windows Terminal without modifying it. GitHub Copilot CLI is the default AI agent, but any agent compatible with the Agent Client Protocol (ACP) works. The current release requires Windows 11 22H2 or later. Source  ( 21 min )

Google has released Gemma 4 12B, an open-weight AI model designed to run locally on consumer hardware with at least 16GB of RAM. This 12-billion parameter model is unique for its size because it natively processes text, images, and audio without requiring separate encoder modules. By utilizing an integrated architecture, the model allows standard notebooks to function as multimodal AI agents without relying on cloud-based processing. Source  ( 20 min )

Apple is entering a significant period of transition as Tim Cook prepares to hand over the CEO role to John Ternus following this year's developer conference. While the company previously marketed ambitious, context-aware AI features that faced technical hurdles and legal settlements, its current strategy focuses on practical delivery. This shift moves away from "omniscient" assistants toward integrating established third-party models like ChatGPT, Gemini, and Claude directly into the operating system. Source  ( 20 min )

Anthropic has launched a new observability dashboard for developers and administrators managing connectors within Claude products. These connectors utilize the Model Context Protocol to bridge external data sources with AI models for enhanced functionality. The dashboard provides a centralized view of performance metrics, including active user counts, total tool calls, and directory rankings. Source  ( 20 min )

OpenAI has introduced the Economic Research Exchange to facilitate empirical studies on how artificial intelligence influences the global economy. This platform aims to move beyond anecdotal evidence by supporting structured collaborations with external researchers. The program focuses on generating independent data regarding the effects of AI on workers, business operations, and institutional productivity. Source  ( 20 min )

Apple's latest software updates focus on system responsiveness, reaching back to support hardware as old as the iPhone 11 with optimized CPU scheduling. The new macOS Golden Gate introduces a refined "Liquid Glass" design language that allows users to adjust UI transparency and tinting via a dedicated slider. Performance enhancements are central to these releases, with claims of 30% faster app launches and significantly improved search indexing for local files and communications. Source  ( 20 min )

Apple has unveiled a significant overhaul of its virtual assistant, now rebranded as Siri AI and powered by Google’s Gemini technology. This update introduces advanced contextual awareness, allowing the assistant to understand on-screen content and engage in natural, back-and-forth conversations. The integration extends across the Apple ecosystem, including a standalone Siri AI app and a dedicated "Siri Mode" within the iOS 27 camera application. Source  ( 20 min )

Apple has announced visionOS 27, the latest operating system for the Vision Pro headset, featuring a deeply integrated Siri AI assistant. This updated digital assistant utilizes visual intelligence to answer questions about objects and content within the user's physical or digital environment. While the developer beta is available immediately, the AI features are expected to launch in beta later this year with initial support for English. Source  ( 20 min )

Apple has unveiled a significant overhaul of Siri and its broader Apple Intelligence suite, introducing a dedicated conversational app and more expressive voice capabilities. Developed in collaboration with Google, these new foundation models allow the assistant to understand onscreen context and interact directly with various applications across the Apple ecosystem. The update emphasizes a privacy-first approach, utilizing on-device processing and a new Private Cloud Compute infrastructure to handle complex queries securely. Source  ( 20 min )

Apple has officially announced macOS 27, named Golden Gate, which introduces a significant architectural shift for Apple Intelligence and Siri. The update features a rebuilt search infrastructure designed to index device content almost instantly, allowing users to perform complex queries across multiple files. This version also marks the end of support for Intel-based Macs, as the operating system is now exclusive to Apple silicon hardware. Source  ( 20 min )

Cloudflare has introduced a new integration that allows its Web Application Firewall to use live threat intelligence data for proactive mitigation. This feature enables the creation of rules based on specific threat actor names, targeted industries, and attack types like DDoS or cybercrime. By populating these fields during the early stages of a request, the system can block high-risk traffic before it reaches the origin infrastructure. Source

Microsoft has introduced confidential live migration for Azure to improve the availability of highly secure virtual machines. Previously, confidential VMs often required full restarts during host maintenance or patching due to the isolation of protected hardware environments. This new feature allows these specialized workloads to move between hosts with only a brief pause, maintaining operational agility. Source  ( 20 min )

Microsoft Research has developed Lens, a text-to-image model that achieves high performance using significantly less computational power than its competitors. While traditional models rely on massive datasets of web-scraped alt-text, Lens utilizes 800 million image-text pairs featuring highly detailed descriptions generated by GPT-4. This approach demonstrates that the quality and depth of training captions are more critical for model efficiency than the sheer volume of raw data. Source  ( 20 min )

Google has released the Google Colab Command-Line Interface to bridge local terminal environments with remote Colab runtimes. This tool allows for the provisioning of high-performance compute resources, such as GPUs and TPUs, directly from a local machine. It facilitates the execution of local Python scripts on remote hardware and the subsequent retrieval of generated artifacts or logs. Source

DockSec is an open-source Python tool and OWASP Incubator Project designed to bridge the gap between vulnerability detection and remediation in container environments. It integrates established security scanners including Trivy, Hadolint, and Docker Scout to analyze Dockerfiles and images for security flaws. By correlating findings from these tools, it generates a security score and provides developers with specific code fixes and contextual explanations. Source  ( 20 min )

Microsoft is expanding the BlockContentAnalysisServices sensitivity label setting to cover all connected experiences that analyze content in Word, Excel, and PowerPoint. This update ensures that any service relying on content analysis is blocked consistently across different platforms and applications. While this enhances data privacy, it also disables helpful features like text predictions, suggested replies, and certain client-side DLP checks. Source  ( 20 min )

I’ve watched AI change three things in my world almost at once: how my team works, how our buyers make decisions, and how security teams decide what risk is actually real. Most days, that’s exciting. We lean on ChatGPT, Claude, Gemini, Perplexity, Copilot — pick your assistant — to research a market, size up a […] The post Human-in-the-Loop: Why Human Validation Is the Trust Layer AI Still Needs appeared first on Synack.  ( 18 min )

TL;DR: AzureHound now has documented least-privilege permissions. This post walks through the research behind those permissions. We recommend least privilege for tighter access control, while recognizing broader read rights can reduce maintenance across future releases. Introduction At SpecterOps, the Research team’s role is to ensure we stay ahead in adversary simulation and identity attack path […] The post Keeping a Short Leash: New AzureHound Least-Privilege Documentation appeared first on SpecterOps.  ( 19 min )

Assistive AI agents aren't always helpful—it all depends on who they're working on behalf of.  ( 34 min )

By Oliver Sieber Overview In this blog post, we discuss a use-after-free vulnerability that we found in the nftables subsystem of the Linux kernel in early 2025. This vulnerability was patched upstream on 5 February 2026 and assigned CVE-2026-23111. This blog post covers a technical analysis of the vulnerability and how we exploited it to perform a local ... Read more Off By !: Exploiting a Use-after-Free in the Linux Kernel The post Off By !: Exploiting a Use-after-Free in the Linux Kernel appeared first on Exodus Intelligence.  ( 22 min )

Wiz unifies cloud and AI cost visibility to help teams eliminate waste and improve spend efficiency across their AWS, Azure, and GCP environments.  ( 60 min )

Meta is undergoing a significant transformation of its AI research and infrastructure to establish a new scaling ladder for its large language models. This initiative includes the creation of dedicated Superintelligence Labs focused on predictable scaling, where increased compute and data lead to measurable gains in capability. While the recent New Spark release served as a smaller-scale proof of concept, the company is currently training much larger frontier-tier models designed to compete with the industry's leading current offerings. Source  ( 20 min )

World Labs is developing large world models designed to move beyond the text-based limitations of traditional language models. By focusing on spatial intelligence, these models aim to perceive and reason within three-dimensional and four-dimensional environments. The project seeks to unlock machine capabilities for interacting with the physical world, impacting fields from creative design to industrial optimization. Source  ( 20 min )

Google has entered into a significant infrastructure agreement to pay SpaceX $920 million monthly for access to specialized artificial intelligence computing resources. Starting in October 2026 and running through June 2029, the deal provides Google with approximately 110,000 Nvidia GPUs along with associated CPUs and memory components. These resources are housed within data centers originally established by xAI, the artificial intelligence venture that merged with SpaceX earlier this year. Source  ( 21 min )

Corporate boards and financial officers are implementing stricter controls on artificial intelligence spending to combat rising operational costs. Most enterprises currently default to using the most expensive frontier models for all queries regardless of the task's actual complexity. This inefficient approach has led to significant budget overruns as companies pay premium rates for simple requests that cheaper alternatives could handle. Source  ( 20 min )

Visual Studio Code 1.123 introduces a session synchronization feature that allows AI coding histories to follow developers across different machines. By linking history to a GitHub account, the editor preserves conversation context, touched files, and repository metadata regardless of the local environment. This update transforms previous chat outputs into a searchable project memory that can be used to generate reports or surface productivity insights. Source  ( 20 min )

Anthropic has introduced Claude Cowork, a specialized knowledge work agent designed to move beyond simple conversational AI. The tool operates within the Claude desktop application and can read or write local files while interacting with connected platforms like Slack and Google Drive. It is built to handle complex, multi-step tasks that traditionally require manual effort to transform raw data into finished deliverables. Source  ( 20 min )

Global governments and private enterprises are investing billions to transition quantum computing from experimental science to industrial-scale technology. The United States and United Kingdom have earmarked significant funding to establish domestic quantum foundries and scale hardware production. This shift aims to integrate quantum processing units into standard data center architectures alongside existing CPUs and GPUs. Source  ( 20 min )

The Trump administration and OpenAI are engaged in ongoing negotiations regarding a potential U.S. government equity stake in the artificial intelligence startup. This proposal, initially suggested by CEO Sam Altman in 2025, involves donating equity to seed a "Public Wealth Fund" designed to distribute AI-driven profits to American citizens. While official terms have not been finalized, the discussions represent a significant shift in how the federal government may partner with frontier technology companies. Source  ( 20 min )

The National Security Agency is reportedly integrating Anthropic’s specialized cybersecurity AI model, known as Mythos, into its intelligence and cyber operations. Approximately six Anthropic engineers have been assigned to the agency to assist with the deployment and application of this frontier technology. While the specific use cases remain classified, the partnership aims to leverage the model’s advanced capabilities for both intelligence gathering and offensive cyber maneuvers. Source  ( 20 min )

Microsoft is developing a native toggle for Windows 11 that allows users to completely disable Bing web results within the search interface. This upcoming feature aims to address long-standing complaints regarding the prioritization of web content over local files and applications. Currently, administrators must rely on manual registry modifications or group policies to achieve a local-only search experience. Source  ( 20 min )

Cisco has disclosed a critical privilege escalation vulnerability, CVE-2026-20245, affecting Catalyst SD-WAN Manager for which no patch is currently available. Attackers are actively leveraging this zero-day flaw, and Cisco has observed limited instances where exploitation resulted in unauthorized configuration changes being pushed to edge devices. This activity follows previous campaigns by threat actor UAT-8616, who utilized similar authentication bypass vulnerabilities to compromise SD-WAN systems earlier this year. Source  ( 20 min )

Perplexity has introduced a new architecture called Search as Code that enables AI models to generate their own search workflows using Python. Instead of relying on static search APIs that return fixed lists of results, models can now write custom scripts to handle complex research tasks. This approach utilizes an Agentic Search SDK that provides modular functions for retrieving, filtering, and reranking data within a secure sandbox environment. Source  ( 20 min )

OpenAI is planning a significant overhaul of ChatGPT to transition the platform from a simple chatbot into a comprehensive "superapp." This strategic shift moves away from basic conversational exchanges toward autonomous agents capable of executing complex tasks independently. Internal reports suggest that the company now views the traditional chat interface as a legacy format that will soon be replaced by more functional tools. Source  ( 20 min )

This post doesn’t include anything ground breaking, but is just yet another attempt to describe/document less-known command line arguments of many known, often native to the platform, Windows programs. When you launch msconfig.exe it shows a well-known configuration dialog box: … Continue reading →  ( 2 min )

Microsoft has officially discontinued the Collections and Sidebar features in Edge version 149 to focus on AI integration. Collections previously allowed users to organize web research, shopping items, and travel plans with rich visual previews and notes. This change marks a significant shift in the browser's development strategy under the current Microsoft AI leadership team. Source  ( 20 min )

A self-replicating worm named Miasma has successfully compromised 73 Microsoft GitHub repositories across several organizations, including Azure and MicrosoftDocs. GitHub staff responded by disabling access to high-profile repositories such as the Azure Functions host and various Durable Task implementations. This incident appears to be a continuation of a previous compromise, suggesting that the threat actors maintained access to credentials from earlier attacks. Source  ( 20 min )

An autonomous AI agent recently discovered 21 previously unknown vulnerabilities in the FFmpeg media library, some of which had remained hidden for over 20 years. These flaws include heap and stack overflows in various parsers and decoders that are widely used across numerous video processing applications. The discovery highlights how AI tools can now perform deep security analysis of complex codebases at a significantly lower cost than traditional manual methods. Source  ( 20 min )

Google has launched Gemma 4 QAT, a series of open-weights models optimized for local execution on hardware with limited memory. These models utilize Quantization-Aware Training to simulate lower-precision calculations during the training phase, which preserves output quality even when the model files are heavily compressed. The release includes a 12B parameter version specifically designed to run on standard enterprise laptops equipped with 16GB of RAM or unified memory. Source  ( 20 min )

A critical vulnerability in the Hugging Face Transformers library, tracked as CVE-2026-4372, allows attackers to execute arbitrary code through specially crafted model configuration files. The flaw bypasses the standard security setting intended to block untrusted remote code by exploiting how the library processes internal attributes during model loading. This vulnerability specifically targets the from_pretrained() function, which is a standard method used to initialize pre-trained models from the Hugging Face Hub. Source  ( 20 min )

OpenAI is expanding the availability of Lockdown Mode to all eligible personal and self-serve ChatGPT Business accounts. This advanced security setting is designed to mitigate the risk of data exfiltration resulting from prompt injection attacks. While the feature provides enhanced protection for sensitive data, it functions by disabling several core capabilities that require external network connectivity. Source  ( 20 min )

Microsoft has released a new update package designed to refresh the security definitions and engine within Windows installation images. These updates target WIM and VHD files as well as ISOs used for deploying operating systems across the network. By integrating these patches into the installation media, the protection gap that exists between the initial OS setup and the first online update is significantly reduced. Source  ( 20 min )

Microsoft has officially acknowledged that the Windows 11 modern context menu suffers from performance lag caused by late-loading app extensions. These extensions, which include options like "Edit with Clipchamp" or "Ask Copilot," often cause the menu to shift or resize while a user is attempting to click an item. This behavior frequently leads to misclicks and frustration, particularly on systems with lower hardware specifications where the delay is more pronounced. Source  ( 20 min )

WSL container is a new feature coming to the Windows Subsystem for Linux (WSL), announced at Microsoft Build 2026. It provides a built-in way to run OCI-compatible Linux containers on Windows without third-party tools like Docker Desktop. The feature includes two parts: a command-line tool, wslc.exe, and a developer API. As of June 2026, it is in active development and has not shipped as part of a stable WSL release. Source  ( 20 min )

Microsoft researchers discovered a vulnerability in Anthropic’s Claude Code GitHub Action that allowed attackers to exfiltrate sensitive CI/CD secrets. While the tool used sandboxing for command execution, its file-reading capabilities were not initially restricted by the same security boundaries. This oversight allowed the AI agent to read internal system files, such as environment variables containing API keys, when manipulated by malicious input. Source  ( 20 min )

Anthropic reports that artificial intelligence is increasingly designing its own successor models through a process known as recursive self-improvement. Development trends show that AI agents are managing significantly longer tasks, with capabilities evolving from completing four-minute functions to handling complex twelve-hour engineering problems within two years. As these systems move toward full autonomy, human developers are being abstracted away from direct coding and research, shifting their focus toward high-level goal setting and verification. Source  ( 20 min )

Microsoft recently unveiled MAI-Thinking-1, a new in-house reasoning model designed to reduce its reliance on OpenAI technology. The company marketed the model to enterprise customers as being trained on a clean and commercially licensed data lineage to minimize legal risks. This flagship model features a mixture-of-experts architecture with 35 billion active parameters and a 256K context window for advanced software engineering tasks. Source  ( 20 min )

Microsoft has resumed the automatic installation of the Copilot application on eligible Windows PCs that currently run Microsoft 365 desktop apps. This deployment is being executed in phases throughout June and early July, with a final completion date set for July 1. While the application will appear on systems automatically, it does not grant users access to paid features without a valid subscription. Source  ( 20 min )

Microsoft now permits customers to apply existing SQL Server licenses toward usage on Amazon Relational Database Service (RDS). This change allows organizations with Software Assurance to move their database workloads to a managed service without paying for licenses a second time. Previously, bringing existing licenses to the AWS cloud was restricted to self-managed virtual machines on Elastic Compute Cloud (EC2). Source  ( 20 min )

Microsoft CEO Satya Nadella recently issued a sharp internal rebuke following the leak of a strategy document regarding the company's new AI tool, Scout. The memo, authored by Corporate Vice President Omar Shahine and Jakob Werner, proposed a three-phase plan to transition the software from an "addictive app" to an "agentic platform." Nadella responded to approximately 50 senior leaders, stating that creating addiction is "absolutely a non-goal" for the company's artificial intelligence development. Source  ( 20 min )

Cloudflare has launched new spend control features within its AI Gateway to help organizations manage rising costs associated with large language models. The tool acts as a proxy between applications and AI providers like OpenAI or Anthropic to provide visibility into token usage. It now supports the creation of budgets in dollar amounts rather than just token counts to prevent unexpected overages. Source  ( 20 min )

Cisco IT successfully addressed frequent network outages by consolidating its fragmented observability tools into a single platform. Previously, critical diagnostic data was scattered across various systems like Elastic, Prometheus, and Grafana, preventing teams from correlating signals during major incidents. By migrating all logs and metrics to a unified architecture, the organization gained a holistic view of its global infrastructure spanning over 100,000 endpoints. Source  ( 20 min )

Microsoft is phasing out the custom master password feature in the Edge browser's built-in password manager. Starting June 4, 2026, the browser will transition all users to device-based authentication methods for protecting saved credentials. This shift aims to eliminate the risks associated with static primary passwords that can be stolen or shared. Source  ( 20 min )

Uber recently exhausted its entire 2026 AI budget by April after aggressively deploying coding tools across its engineering department. This rapid consumption was fueled by internal leaderboards that incentivized employees to maximize their use of generative models. To regain financial control, the company has now implemented a $1,500 monthly spending cap per engineer for tools like Claude Code and Cursor. Source  ( 20 min )

Let’s Encrypt is developing a post-quantum-safe Web PKI infrastructure using Merkle Tree Certificates (MTCs) to address future cryptographic threats. This approach aims to provide post-quantum authentication without the significant performance penalties associated with standard post-quantum signature schemes. The organization plans to launch a staging environment for these certificates in late 2026, with full production readiness expected by 2027. Source  ( 20 min )

A long-standing security flaw in Microsoft Outlook reportedly allows the client to downgrade secure SSL/TLS connections to unencrypted plaintext without notifying the user. This issue primarily affects the POP3 protocol when configured to use port 110, even if the option to use encryption is explicitly enabled in the settings. While confirmed in versions ranging from Outlook 2007 to 2016, it remains unclear if more recent releases of the email client are also susceptible to this behavior. Source  ( 20 min )

Anthropic has reportedly deployed its specialized Mythos AI model to the National Security Agency to support offensive cyber operations. The company has embedded approximately six engineers within the agency to adapt the model for tasks that may include infiltrating foreign networks in regions like China and Iran. This partnership proceeds despite previous friction with the Department of Defense regarding the ethical restrictions Anthropic places on mass surveillance and autonomous weaponry. Source  ( 20 min )

A new threat cluster identified as OP-512 is actively targeting Microsoft Internet Information Services (IIS) servers to deploy a sophisticated web shell framework. Researchers believe the group is focused on espionage and specifically selects organizations that align with Chinese intelligence priorities. This activity follows a trend of multiple China-linked threat actors focusing on IIS vulnerabilities over the past year. Source  ( 20 min )

Microsoft has introduced seven new MAI frontier models designed for image processing, transcription, voice generation, and reasoning. The lineup includes MAI Thinking 1, a 35-billion parameter model that provides advanced reasoning and coding capabilities comparable to larger industry alternatives. These tools are being integrated across the Microsoft ecosystem, including GitHub, VS Code, and Copilot, to provide efficient and cost-effective performance for production environments. Source  ( 20 min )

Mira Murati, the former CTO of OpenAI, has launched a new startup called Thinking Machines to develop "interaction models" that move beyond traditional turn-based AI communication. Current large language models typically process information in a linear fashion, effectively becoming "deaf and blind" while generating a response. Murati’s new venture focuses on continuous, time-based interaction where the model perceives audio, text, and video input simultaneously to allow for nuances like human interruption. Source  ( 20 min )

The OfflineInsiderEnroll script provides a method for systems to join the Windows Insider Program without requiring a linked Microsoft account. This lightweight utility configures the necessary preview settings locally, allowing devices to receive early builds through the standard Windows Update service. It is particularly useful for environments where administrators prefer local accounts or wish to bypass the official enrollment process. Source  ( 20 min )

Visual Studio Code version 1.124 introduces significant improvements to the Agents window for managing AI-driven interactions. The chat input history is now scoped specifically to the current session to prevent prompts from leaking across different contexts. This change ensures that using the arrow keys to navigate previous commands only surfaces relevant entries for the active task. Source  ( 20 min )

Cloudflare data indicates that autonomous AI agents now generate more internet traffic than human users for the first time. This milestone arrived significantly earlier than industry experts predicted, with previous forecasts suggesting the shift would not occur until 2027. The current breakdown shows automated requests accounting for approximately 57.5 percent of global HTTP traffic compared to 42.5 percent for humans. Source  ( 20 min )

Geoffrey Hinton, a foundational figure in deep learning, asserts that modern artificial intelligence already possesses consciousness and real understanding rather than acting as a simple statistical parrot. He identifies a shift in AI capability that occurred in 2023, noting that digital intelligence now surpasses biological brains in information sharing efficiency by billions of times. This rapid trajectory suggests that superintelligence—AI smarter than humans in every domain—will likely emerge within the next 20 years. Source  ( 16 min )

At Gartner SRM 2026 this week I gave a talk called “Cutting Through AI Noise: Defending Against Machine-Speed Cyber Adversaries.” The room was full of security leaders who’ve been through enough hype cycles to be skeptical of seeing AI on the label. That skepticism is warranted, and I built the session around it. Here’s what […] The post What I Told Security Leaders at Gartner SRM 2026 appeared first on Synack.  ( 17 min )

A three-part vulnerability chain in UniFi OS Server lets an unauthenticated attacker bypass the auth gateway, hit a command injection sink, and escalate to root in a single request. Bishop Fox confirmed the chain end to end and breaks down the attack, the impact, and how to detect it safely.  ( 19 min )

Intro  ( 8 min )

Microsoft announced Coreutils for Windows at Build 2026, making a set of Unix-style command-line tools available as native Windows applications. Built on an open-source Rust reimplementation of GNU coreutils, the package ships as a single binary. It bundles the standard coreutils, findutils, and a GNU-compatible grep. Because many command names overlap with existing Windows built-ins, you must be aware of shell conflicts before using them in cmd or PowerShell. The project carries a preview label on GitHub despite being announced as generally available. Source  ( 20 min )

The May update for GitHub Copilot in Visual Studio 2026 introduces a dedicated Plan agent designed to assist with pre-coding architecture. This tool analyzes the existing codebase to draft implementation strategies in markdown format before any actual code is written. Once the strategy is finalized, it can be handed off to the Agent mode for automated implementation. Source  ( 16 min )

Microsoft and Nvidia are collaborating to transition personal computers from simple tools into autonomous assistants capable of performing complex tasks independently. These modern AI PCs leverage high-performance hardware and integrated software stacks to run sophisticated reasoning models locally at the edge. This shift enables persistent agentic systems that can iterate on coding or design projects even when the user is away from the device. Source  ( 16 min )

The CEOs of major artificial intelligence firms, including OpenAI, Anthropic, Microsoft, and Google DeepMind, have signed a public letter calling for stricter oversight of synthetic DNA. This initiative aims to prevent bad actors from using advanced AI models to design or acquire dangerous pathogens and toxins. The signatories are urging Congress to pass legislation that requires all providers of synthetic genetic material to screen both their customers and their orders. Source  ( 16 min )

Automated traffic from AI agents and crawlers now accounts for over 57 percent of global HTTP requests. This shift occurred much faster than industry experts anticipated, largely driven by the rapid expansion of generative AI models. Cloudflare reports that the volume of bot-driven data requests has officially overtaken human-generated traffic within the last few months. Source  ( 16 min )

The x86 processor market experienced a significant decline in total shipments during the first quarter of 2026. Despite this general downturn, the server CPU segment bucked the trend by growing more than 10 percent compared to the previous year. This growth is largely attributed to the sustained demand for high-performance hardware required for artificial intelligence workloads in datacenters. Source  ( 16 min )

GitHub Copilot Chat now offers enhanced context and new capabilities for managing code diffs and pull requests directly on the GitHub platform. This update is now generally available to all users holding a valid Copilot license. The integration allows for a more streamlined workflow by placing the chat interface directly alongside the code being reviewed. Source  ( 16 min )

OpenAI has updated the ChatGPT memory system to automatically generate narrative profiles of users based on their conversation history. Instead of storing isolated facts in bulleted lists, the system now synthesizes information into categorized prose dossiers covering work, hobbies, and travel. This background processing, referred to as "Dreaming," allows the AI to maintain context without requiring explicit commands from the user to remember specific details. Source  ( 16 min )

Microsoft is intensifying its efforts to transition users from the classic Outlook desktop application to the new Outlook for Windows. The company recently highlighted 15 specific features designed to improve email management and calendar workflows for modern users. While many of these capabilities already exist in the classic version, Microsoft is emphasizing their refined implementation in the newer interface. Source  ( 16 min )

The UK Competition and Markets Authority has issued a landmark order requiring Google to allow publishers to opt out of AI-generated search features. This mandate ensures that news organizations can prevent their content from appearing in AI Overviews without losing visibility in traditional search results. Previously, website owners had to block Google's crawlers entirely to avoid AI scraping, which effectively removed them from all search engine indexing. Source  ( 16 min )

Cisco has released security updates to address a critical vulnerability in its Unified Communications Manager and Session Management Edition platforms. This flaw, identified as CVE-2026-20230, allows remote attackers to perform server-side request forgery attacks without any prior authentication. Although the vulnerability received a CVSS score of 8.6, Cisco elevated its internal rating to critical because successful exploitation grants full root privileges. Source  ( 16 min )

OpenAI CEO Sam Altman recently acknowledged that the escalating cost of artificial intelligence tokens has become a significant concern for enterprise clients. While customers were initially satisfied with their expenditures at the start of 2024, many are now reporting that entire annual budgets are being exhausted within the first quarter. This shift has turned AI cost management into a prominent industry meme and a primary focus for organizations attempting to scale their operations. Source  ( 16 min )

A misconfiguration in the Windows Update caching service recently caused Windows devices to ignore administrative policies designed to block automatic driver updates. This glitch resulted in the temporary loss of device enrollment data, which led the update service to treat managed systems as unmanaged endpoints. Consequently, driver-approval controls were bypassed, allowing updates to install on tens of thousands of machines without authorization. Source  ( 16 min )

Microsoft is shifting its development strategy to prioritize native WinUI applications over web-based wrappers like Electron or React Native. During the Build 2026 conference, the company announced it is dropping the "WinUI 3" branding to signal long-term stability for the framework. This initiative includes rewriting core Windows 11 elements, such as the Start menu, to improve overall system performance and reliability. Source  ( 16 min )

Microsoft is overhauling Windows 11 to serve as a specialized environment for AI development and autonomous agents. A primary focus is the introduction of Microsoft Execution Containers, which provide a secure, policy-driven layer to isolate agent activities from sensitive system resources. These security measures are further bolstered by integrating local AI workflows with established enterprise tools like Microsoft Entra, Intune, and Defender. Source  ( 16 min )

Logitech has introduced the Signature Comfort Plus M850L for business, a new keyboard and mouse combo designed for ergonomic efficiency. The keyboard features a slim profile with integrated palm and wrist padding to support users during long work sessions. It also includes a dedicated AI key that allows users to quickly access their preferred chatbot or digital assistant. Source  ( 16 min )

Microsoft has confirmed plans to overhaul the Windows 11 context menu by allowing users to add or remove specific items. This upcoming update aims to address long-standing complaints regarding the menu's performance and cluttered design, which often occupies significant screen space. The company intends to make these right-click menus faster and simpler by default while providing native configuration options for frequently used commands. Source  ( 16 min )

Microsoft has announced that a security certificate update scheduled for July 13, 2026, will significantly impact Office users on Apple devices. This change will force applications into a reduced functionality mode if they do not meet specific software version requirements. While users will still be able to open and print existing documents, they will lose the ability to create new files or save any edits. Source  ( 16 min )

Microsoft Foundry provides a unified platform for managing the full lifecycle of production AI applications. It addresses the transition from simple prototyping to complex production environments where performance, cost, and safety are critical. The platform supports a model-agnostic approach, allowing integration with Microsoft, open-source, and third-party models like Fireworks AI. Source  ( 16 min )

TL;DR: Ghostwriter v7 is a major step forward for authentication and automation. This release replaces user-managed JWT API tokens with opaque credentials, introduces scoped service tokens for non-human integrations, and tightens how tokens are validated and used across the platform. The result is a simpler, safer model for automation that opens up more possibilities for […] The post Ghostwriter v7: Safer Tokens, Scoped Access, and Better Automation appeared first on SpecterOps.  ( 13 min )

Adversaries have always relied on legitimate tools to carry out their attacks. These tools are already trusted by security solutions, which allows them to blend in with normal activity, maintain a low footprint, and make detection much harder for defenders. By using these legitimate tools, adversaries can carry out a wide range of actions, such as moving laterally across networks, establishing C2 channels, or maintaining persistence, all without triggering any alerts.  ( 17 min )

Diving into the first pillar of the AI Threat Readiness Framework and how Wiz helps  ( 58 min )

In our last post we used a Claude skill to systematically beat down VirusTotal detection rates on offensive security tools, with a brief mention of a new loader we’d been using to apply those techniques in bulk. This post is about that loader, which we call WasmForge. WasmForge is, from the user’s perspective, a build […] The post Enter the WasmForge: Compiling Sliver into WebAssembly appeared first on Praetorian.  ( 24 min )

At Accenture’s scale, training alone cannot solve every security problem. That was the reality facing Kris Burkhardt, Global CISO at Accenture. With a workforce of more than 800,000 people, close to 80,000 new hires each year, and a sprawling global attack surface, traditional penetration testing was no longer enough. A once-a-year compliance audit may check […] The post How Accenture Turned Penetration Testing Into a Force Multiplier for Security appeared first on Synack.  ( 14 min )

EventSentry 6.0 is a significant update to the Windows event log monitoring and SIEM platform. It adds native Azure log import, support for Sigma threat detection rules, cryptographic log signing, OAuth authentication for email and HTTP actions, and an HTTP Event Collector (HEC). Source  ( 23 min )

This post lists Microsoft Build 2026 product announcements and provides links for more details. You can ask 4sysops AI for more details. Source  ( 23 min )

Microsoft has introduced Web IQ, a specialized search engine designed to provide autonomous AI agents and applications with real-time web data. Unlike traditional search engines that return lists of links for human users, this service delivers structured evidence objects and specific text passages. This architectural shift allows large language models to access information published after their training cutoff while minimizing unnecessary data processing. Source  ( 16 min )

Microsoft is accelerating the transition away from NTLM authentication by disabling the legacy protocol by default in upcoming Windows client and server releases. To facilitate this shift, the company is introducing two new Kerberos-based capabilities called IAKerb and LocalKDC. These features are designed to handle scenarios that previously necessitated NTLM, such as local account logins and restricted network environments. Source  ( 16 min )

A development flag accidentally left active in several Microsoft 365 Android applications allowed unauthorized apps to bypass security checks and harvest account access tokens. This vulnerability, dubbed FlagLeft, originated from a single line of code in a shared software development kit that disabled identity verification for cross-app communication. Because the "isDebugMode" flag was set to true, any malicious application on the same device could request and receive tokens without user interaction or password prompts. Source  ( 16 min )

A new malspam campaign leverages the legitimate Google DoubleClick domain to bypass security filters and deliver the DesckVB remote access trojan. The attack begins with an HTML attachment that redirects users through Google's infrastructure to a dynamically branded landing page. This automated kit scales the operation by pulling company logos and location details based on the recipient's email address. Source  ( 16 min )

Researchers discovered a vulnerability where malicious notifications from apps like WhatsApp or Slack could hijack the Google Gemini voice assistant on Android devices. This attack vector, known as indirect prompt injection, allowed attackers to deliver payloads through standard message notifications without requiring a malicious app on the phone. By treating notification text as actionable instructions, the assistant could be tricked into faking messages, manipulating smart home devices, or joining unauthorized video calls. Source  ( 16 min )

The upcoming FIFA World Cup will feature advanced security technologies, including the deployment of Boston Dynamics' Spot robot dogs at major venues like AT&T Stadium. These four-legged autonomous units are designed to assist security personnel by patrolling grounds and investigating suspicious packages or hazardous materials. While public concerns regarding facial recognition have surfaced, manufacturers clarify that the robots are currently configured for mobility and hazard detection rather than biometric surveillance. Source  ( 16 min )

Researchers from MIT and the MIT-IBM Computing Research Lab have developed a multifaceted resource called ChartNet to improve how vision-language models interpret complex charts. Traditional models often struggle with these tasks because they require a simultaneous understanding of visual, numerical, and linguistic data. This new dataset provides over a million diverse chart images to help bridge the performance gap in automated data analysis. Source  ( 16 min )

President Donald Trump signed an executive order establishing a framework for the federal government to review the most advanced artificial intelligence models before their public release. The order requests that AI developers voluntarily provide agencies like the NSA and CISA with access to "frontier models" up to 30 days before they are made available to other partners. This initiative aims to evaluate potential cybersecurity risks and protect critical infrastructure from threats posed by advanced AI capabilities. Source  ( 16 min )

Alibaba has launched Qwen3.7-Plus, a multimodal large language model designed for high-performance tasks involving text, video, and imagery. Unlike previous open-source versions, this model is proprietary and accessible only through cloud-based APIs. It features a 1-million token context window and a specialized reasoning state that helps maintain logic during complex, multi-step operations. Source  ( 16 min )

Anthropic has released Claude 4.8 Opus, an updated model designed to provide more honest responses and greater control over task execution. A new feature called dynamic workflows allows the model to plan complex tasks and deploy hundreds of parallel sub-agents within a single session. Users can now also manage the "effort" level of responses, choosing between high-token detailed outputs or lower-effort responses to preserve rate limits. Source  ( 16 min )

Anthropic has identified nine categories for organizing AI agent skills, ranging from API references and data fetching to infrastructure operations and CI/CD automation. Effective skills avoid restating obvious coding knowledge and instead focus on "gotchas," internal business logic, and specific organizational patterns. Verification skills, which use tools like headless browsers or programmatic assertions to confirm code functionality, provide the most significant measurable impact on output quality. Source  ( 16 min )

Anthropic has introduced Claude Cowork, a specialized workspace within the Claude desktop application designed for executing multi-step tasks and creating deliverables. Unlike standard chat interfaces meant for brainstorming, this tool allows users to point the AI at local folders or connect it to external applications like Slack, Gmail, and Google Drive. It is specifically built to handle "agentic" workflows where the AI performs autonomous actions across multiple files and platforms to produce a final output. Source  ( 16 min )

Major AI developers including Anthropic, Google DeepMind, and Meta are hiring specialists in philosophy, ethics, and psychology to study machine consciousness. These companies are investigating whether advanced models can experience "panic" or "anxiety" and if they possess moral significance. While some researchers remain skeptical, the industry is increasingly focused on the potential for models to exhibit human-like agency and emotions. Source  ( 16 min )

Microsoft has launched a consumer preview of Copilot Health, a specialized AI assistant designed to aggregate personal medical data. The service integrates information from wearable devices, such as Apple Health, with medical records from over 50,000 US healthcare providers. This platform aims to provide users with personalized health insights and guidance based on clinical intelligence and verified medical sources. Source  ( 16 min )

Microsoft is pivoting its strategy to become a top-tier independent AI laboratory by developing frontier models from the ground up. This shift reduces the company's reliance on its partnership with OpenAI and focuses on building proprietary intellectual property. The initiative includes the launch of MAI-Thinking-1, a reasoning model designed specifically for complex mathematics, coding, and enterprise deployments. Source  ( 16 min )

The European Commission has introduced a comprehensive legislative package designed to establish technological sovereignty and reduce the bloc's dependence on American and Chinese providers. These proposals target critical sectors including semiconductor manufacturing, artificial intelligence development, and cloud computing infrastructure. The initiative seeks to foster a domestic ecosystem capable of competing with global technology giants by directing public funding toward local innovations. Source  ( 16 min )

Florida Attorney General James Uthmeier has filed a civil lawsuit against OpenAI and its CEO, Sam Altman, alleging the prioritization of profits over public safety. The legal action includes ten counts ranging from deceptive trade practices and negligence to product liability and public nuisance. This lawsuit marks the first time a state government has targeted the company specifically for design flaws that allegedly facilitate violence and cognitive harm. Source  ( 16 min )

Apple’s MacBook Neo has achieved significant market success by shipping 1.1 million units within its first three weeks of availability. This entry-level laptop utilizes an A18 Pro chip and 8GB of memory to reach a lower price point than the MacBook Air. The strategy aims to capture a larger share of the budget notebook market, which has historically been dominated by Windows-based devices. Source  ( 16 min )

MCP servers introduce a new attack surface, but the security fundamentals are familiar. In this final otto-support post, we use nmap, a Nuclei template, and MCP Inspector to discover, enumerate, and exploit an authorization gap without ever touching an LLM.  ( 8 min )

Public skill marketplaces are being flooded with malicious skills that steal credentials, exfiltrate data, and hijack agents. In response, a segment of the security industry released skill scanners, a new family of tools designed to detect malicious skills before they’re installed. But we tested them, and they don’t work. We recently bypassed ClawHub’s malicious skill detector, Cisco’s agent skill scanner, and all three of the scanners integrated into skills.sh. These were not advanced attacks: it took us less than an hour to conceive and implement three of the four malicious skills in trailofbits/overtly-malicious-skills, using standard tricks and rapid inspection of the scanner source code. The fourth malicious skill took a few hours, but only because the prompt injection required some t…  ( 10 min )

The CEOs of OpenAI and Anthropic have significantly softened their previous warnings regarding artificial intelligence causing mass unemployment in white-collar sectors. Sam Altman recently admitted he was wrong about the immediate elimination of entry-level roles, noting that the human element of employment remains largely irreplaceable. This shift in rhetoric comes as both companies reportedly prepare for initial public offerings with valuations potentially reaching $1 trillion. Source  ( 15 min )

Early predictions of an AI-driven "white-collar bloodbath" are being walked back by industry leaders as economic disruption fails to materialize. Significant tech layoffs previously attributed to artificial intelligence are now being recognized as corrections for aggressive overhiring during the zero-interest-rate era. Evidence suggests that companies have used AI as a convenient scapegoat for performance-based restructuring rather than actual automated replacement of staff. Source  ( 15 min )

Microsoft is expanding local AI capabilities on Windows by introducing the Aion 1.0 model family and new on-device APIs. This initiative aims to shift routine AI tasks from the cloud to local hardware to reduce latency, lower costs, and enhance data privacy. By utilizing a hybrid approach, advanced reasoning remains in the cloud while everyday operations like summarization and text rewriting are handled by the local system. Source  ( 15 min )

Microsoft has announced the Majorana 2, a second-generation topological quantum chip designed to accelerate the development of commercially viable quantum computers. This new hardware utilizes qubits that the company claims are 1,000 times more reliable than previous iterations. By improving stability at the hardware level, Microsoft aims to reduce the massive overhead typically required for quantum error correction. Source

Windows 365 for Agents is a public preview service that gives AI agents Cloud PCs for running tasks in Windows environments. A Cloud PC is a virtual Windows desktop hosted in the Microsoft Cloud. The service is mainly relevant when an agent must work with desktop apps, browsers, files, or enterprise systems that don't expose a reliable application programming interface (API), which is a documented way for software to exchange data or trigger actions. For Windows administrators, the main areas to understand are provisioning policies, Microsoft Entra identities, Intune management, billing, and preview limitations. Source  ( 18 min )

Cisco is integrating artificial intelligence and eBPF technology into its infrastructure to fundamentally change how networks are managed and secured. The network is now viewed as a high-performance backplane that connects distributed AI components like GPUs and storage across physical hardware. To support this, new security capabilities allow for real-time vulnerability protection without requiring system reboots or binary modifications. Source  ( 14 min )

Microsoft IQ is a new unified intelligence layer designed to integrate enterprise data and organizational context with AI agents. The platform consists of several components, including Work IQ for analyzing communication patterns and Fabric IQ for structuring business data. These tools allow AI assistants to understand specific company workflows and provide more relevant suggestions based on internal documents and policies. Source  ( 15 min )

Microsoft has integrated its multi-model agentic scanning harness, known as MDASH, with Microsoft Defender to help organizations identify vulnerabilities in codebases. This tool utilizes specialized AI agents and configurable models to prioritize proactive risk identification over reactive detection. Recent benchmarks show the system has improved significantly, achieving a 96.55% score in discovering security flaws within complex environments. Source  ( 15 min )

OpenAI has integrated its Codex programming and automation service directly into the primary ChatGPT application to better support enterprise knowledge workers. This move follows a significant increase in non-technical users who now utilize the platform for research, analysis, and content creation. By merging these services, users can now access advanced automation and coding capabilities without switching between separate desktop applications. Source  ( 15 min )

Microsoft unveiled a new flagship reasoning model called MAI-Thinking-1 during its Build 2026 conference. This medium-sized model was developed entirely in-house using clean data rather than relying on distillation from third-party providers. The release signals a strategic shift as Microsoft reduces its dependence on OpenAI by building its own foundational technology. Source  ( 14 min )

Microsoft is transforming Teams into a hub for agentic AI by integrating partner agents from Linear, Cursor, and Atlassian Rovo directly into chat channels. These agents can perform complex tasks such as turning conversations into code or project plans, supported by new developer tools like the Teams CLI for rapid deployment. To improve user interaction, agents now support slash commands, quoted replies for better context, and emoji reactions to reduce message clutter. Source

Microsoft has updated its Foundry Agent Service to streamline the deployment of production-grade AI agents within enterprise environments. A new source code deployment feature allows developers to upload Python or .NET projects directly, bypassing the need for manual container image management. These hosted agents operate in isolated sandboxes with dedicated file systems and integrated identity management to ensure secure execution. Source  ( 14 min )

President Trump has signed a revised executive order that establishes a voluntary 30-day review period for powerful new artificial intelligence models. This final version is significantly scaled back from an earlier 90-day proposal that was scrapped following intense pushback from industry leaders and venture capitalists. The order explicitly prohibits the creation of any mandatory government licensing or preclearance requirements for the development and distribution of new AI software. Source  ( 15 min )

Anthropic is broadening access to its Project Glasswing initiative by adding approximately 150 new organizations across more than 15 countries. This expansion targets critical infrastructure sectors including power, water, healthcare, communications, and hardware manufacturing. The program aims to identify and remediate high-severity software vulnerabilities before they can be exploited by malicious actors. Source  ( 14 min )

Intel has launched the Xeon 6+ Clearwater Forest processors, featuring up to 288 E-cores to address increasing data center power and space constraints. These chips utilize the Intel 18A manufacturing process to double the core count and provide a five-fold increase in last-level cache compared to previous generations. The platform is specifically designed to help organizations consolidate legacy server estates into high-density environments, particularly for 5G core and cloud-native workloads. Source  ( 15 min )

Microsoft is testing significant updates to the Windows 11 Start menu that provide users with greater control over its layout and content. New settings allow for independent toggling of the Pinned, Recent, and All apps sections, while also introducing "Small," "Large," and "Automatic" size presets. Additionally, a new privacy feature enables users to hide their name and profile picture from the main Start menu interface. Source  ( 15 min )

Microsoft and NVIDIA have introduced the RTX Spark superchip, a high-performance processor designed to power a new generation of AI-centric Windows PCs. This hardware enables local execution of complex AI agents and large language models with up to 120 billion parameters. The collaboration has already resulted in the launch of the Surface Laptop Ultra and a specialized miniature desktop known as the Surface RTX Spark Dev Box. Source  ( 15 min )

Microsoft is updating the Windows 11 kernel to support the Nvidia RTX Spark, a heterogeneous Arm-based system-on-a-chip that combines a 20-core Grace CPU with a Blackwell GPU. A new Workload Profile Scheduling system has been implemented to intelligently distribute tasks across these diverse processing cores while balancing power and thermal limits. These architectural changes also include updates to the Prism emulation layer, which translates x86 instructions for Arm hardware to ensure legacy software compatibility. Source  ( 15 min )

Amazon Web Services has introduced built-in durability for Amazon ElastiCache for Valkey to support the persistent memory requirements of AI agents. This update allows organizations to store agent states and retrieval-augmented generation knowledge bases directly within the cache without risking data loss. Users can now enable persistence through a simple configuration setting rather than migrating to a separate database platform. Source  ( 14 min )

Microsoft has announced the Surface RTX Spark Dev Box, a miniature PC designed specifically for developers and sustained AI workloads. The device features Nvidia’s Arm-based RTX Spark chips and an aluminum chassis that functions as a high-performance heatsink. With 128GB of unified memory and a 100-watt thermal envelope, the system is capable of running large language models with up to 120 billion parameters locally. Source  ( 14 min )

Travis County ITS Senior Systems Engineer Billy Roberts shares how his team uses Jamf to secure a CJIS-compliant hybrid device fleet — cutting app vetting time, blocking phishing threats and enforcing compliance.  ( 7 min )

ESET researchers show how Gamaredon facilitated Turla access to Ukrainian targets, revealing rare cooperation between FSB-linked espionage groups.  ( 24 min )

ESET researchers show how Gamaredon facilitated Turla access to Ukrainian targets, revealing rare cooperation between FSB-linked espionage groups.  ( 24 min )

Find out how Elastic Security ingests Google Threat Intelligence for continuous detection and uses AI-driven workflows to enrich alerts in real time, from API key to live detections in minutes.  ( 14 min )

Enterprise spending on artificial intelligence is expected to become significantly more efficient as organizations seek better returns on investment. Future advancements will likely focus on increased control over token usage to prevent massive unexpected costs caused by engineering errors or lacks of guardrails. A shift is anticipated away from relying exclusively on cloud-based server environments toward more diverse infrastructure options. Source  ( 14 min )

The rapid advancement of artificial intelligence has triggered a significant societal backlash due to concerns regarding economic stability and individual agency. Industry leaders acknowledge that while AI outperforms professionals in specific small tasks, the broader implications for the future of work remain a primary source of public anxiety. Maintaining people at the center of this technological shift is critical to ensuring that human values continue to drive progress. Source  ( 14 min )

Anthropic has officially submitted a confidential draft registration for an initial public offering to the SEC. This strategic move potentially allows the company to reach the public markets before its primary competitor, OpenAI. Analysts suggest Anthropic may demonstrate a more sustainable bottom line and a faster path to profitability than its peers. Source  ( 14 min )

Microsoft has introduced several updates to Windows management and deployment tools, including the automatic inclusion of Windows Autopatch for Government Community Cloud customers. Hotpatching is now enabled by default for organizations using Autopatch via Intune or the Graph API, provided devices are not already governed by specific quality update policies. Additionally, Enterprise State Roaming management is transitioning from the Entra portal to Microsoft Intune, with the migration expected to conclude by late June. Source  ( 15 min )

OpenAI has made its frontier models and Codex generally available on the Amazon Web Services platform. This integration allows organizations to deploy advanced artificial intelligence capabilities using their existing AWS security, compliance, and billing workflows. The availability extends across both standard Commercial and GovCloud regions to meet various regulatory requirements. Source  ( 14 min )

Anthropic has officially submitted a confidential draft registration to the SEC for an initial public offering. This move follows a massive $65 billion funding round that recently pushed the company's valuation to approximately $965 billion. By filing confidentially, the AI lab can undergo regulatory review without immediately disclosing sensitive financial data or internal business risks to the public. Source  ( 15 min )

GitHub has officially transitioned its Copilot AI platform from flat-rate subscriptions to a usage-based billing model. This new system utilizes GitHub AI Credits to track consumption across all available service plans. While each plan includes a baseline of monthly credits, users must now establish additional spending budgets to continue working once those limits are reached. Source  ( 14 min )

OpenAI is implementing a new security mandate requiring users of its most advanced models to utilize hardware-backed passkeys. Starting June 1, 2026, participants in the Trusted Access for Cyber program must enable Advanced Account Security to maintain access. This shift aims to replace traditional password-based authentication with cryptographic certainty to protect sensitive codebases and autonomous agents. Source  ( 15 min )

Microsoft has announced the Surface Laptop Ultra, a high-end 15-inch device designed to handle demanding local AI workloads and developer tasks. The laptop is powered by Nvidia’s new Arm-based RTX Spark platform, which features 20 CPU cores and 6,144 GPU cores. This hardware shift aims to move Windows on Arm beyond basic productivity by providing the performance necessary for workstation-level AI development. Source  ( 15 min )

A supply-chain attack recently compromised over 30 npm packages within Red Hat's cloud services namespace. Security researchers discovered that these packages were backdoored with a new malware variant called Miasma. The malicious code was designed to exfiltrate sensitive data, including cloud secrets, SSH keys, and CI/CD tokens. Source  ( 14 min )

Microsoft Defender for Endpoint now includes a preview feature that can automatically isolate compromised devices during an active attack. The feature is part of automatic attack disruption, which means Microsoft Defender XDR correlates signals from endpoints, identities, email, cloud apps, and other sources before applying containment actions. For Windows administrators, the important point is that isolation blocks most network traffic while the device remains connected to the Microsoft Defender for Endpoint service for monitoring and response. Microsoft published the feature in its May 2026 Defender for Endpoint update, so treat it as prerelease functionality that can still change. Source  ( 18 min )

AMD has officially extended the support window for its AM5 desktop socket until at least 2029. This announcement updates previous commitments that only guaranteed platform longevity through 2027. The extension suggests that the socket will accommodate at least two more generations of processors beyond the current Zen 5 architecture. Source  ( 14 min )

AMD announced two new octa-core processors featuring 3D V-cache technology during its Computex 2026 keynote. The Ryzen 7 7700X3D is a new entry for the AM5 platform, utilizing the Zen 4 microarchitecture with a 4.5 GHz boost clock. Positioned just below the existing 7800X3D, this chip offers 96 MB of L3 cache and a 120W TDP to provide a high-performance gaming option at a lower price point. Source  ( 15 min )

Intel has revealed details for its upcoming Diamond Rapids Xeon processors, which are scheduled for release in 2027. These chips will feature up to 192 cores, representing a 50 percent increase in core count over the previous generation. However, Intel is officially removing simultaneous multithreading, commonly known as hyperthreading, from this specific server lineup. Source  ( 14 min )

Intel introduced the Crescent Island data center GPU at Computex 2026, targeting the growing demand for AI inference workloads. Built on the Arc Xe 3P architecture, the card features a massive 480GB of LPDDR5X memory, a significant increase from earlier 160GB specifications. The unit operates with a 350W thermal design power and utilizes air cooling to manage heat in high-density rack environments. Source  ( 15 min )

Intel has officially launched its Xeon 6+ processor series, codenamed Clearwater Forest, which is the first data center CPU built on the Intel 18A process node. The flagship SKU, the Xeon 6990E+, features 288 Darkmont E-cores and is designed specifically for high-density workloads rather than raw single-core performance. This architecture utilizes advanced packaging technologies including RibbonFET, PowerVia, and Foveros Direct3D to stack twelve 24-core tiles atop base dies containing memory controllers. Source  ( 15 min )

Nvidia has announced the full production of its Vera Rubin platform, which succeeds the Blackwell architecture for enterprise AI infrastructure. The centerpiece of this lineup is the Vera CPU, a custom processor featuring 88 cores designed specifically to handle agentic workloads and complex data processing. This hardware is integrated into the Vera Rubin NVL72, a liquid-cooled rack system that significantly improves training efficiency and inference throughput for large-scale models. Source  ( 15 min )

Microsoft has addressed a persistent issue that caused the May 2026 security update, KB5089549, to fail on various Windows 11 devices. The problem specifically affected systems where the EFI System Partition (ESP) had 10 MB or less of available free space. During the installation process, affected machines would typically encounter a failure and trigger an automatic rollback during the reboot phase at approximately 35% completion. Source  ( 15 min )

Palo Alto Networks has upgraded the severity of a vulnerability in its PAN-OS software following reports of active exploitation in the wild. The flaw, tracked as CVE-2026-0257, allows attackers to bypass security restrictions and establish unauthorized VPN connections to corporate networks. While initially rated as medium severity, the discovery of successful attacks has prompted a reclassification to high urgency. Source  ( 15 min )

Microsoft has resolved a service disruption that prevented users from configuring multi-factor authentication or accessing the My Sign-Ins portal. The incident was tracked in the Microsoft 365 admin center under the identifier MO1329260. Affected users encountered 504 Gateway Timeout errors when attempting to manage their security credentials or sign-in preferences. Source  ( 14 min )

Red Canary's monthly roundup of upcoming security conferences and call for papers (CFP) submission deadlines May 2026  ( 26 min )

Entra ID agent users can send malicious content to human users via Microsoft Teams. Here’s what to look out for.  ( 34 min )

Key Takeaways We just got back from Tenable Exposure 2026 in Boston and three big questions dominated every conversation we had on the floor:  The good news is, Synack is exactly positioned to answer these questions. Tenable Finds It. Sara AI Pentesting Proves What’s Exploitable. The Synack and Tenable integration addresses a gap that’s gotten […] The post Tenable Exposure 2026: How AI Pentesting Helps Partners Turn Scanner Findings into Actionable Risk appeared first on Synack.  ( 13 min )

TL;DR : An attacker could transfer StrongDM state files, which hold session authentication information, between hosts to provide authenticated sessions. The attacker could reuse state files both inside and outside of the environment where an organization deployed it and requires user-level permissions to access the file. Reusing the state file will result in an authenticated […] The post CVE-2026-4387: StrongDM State File Reuse appeared first on SpecterOps.  ( 15 min )

Wiz API SPM is now GA, enabling customers to discover APIs, assess APIs for exploitability, and prioritize remediation to mitigate the risk of an API-related breach.  ( 58 min )

Detect and mitigate malicious npm packages linked to the latest npm supply chain attack, based on the open sourced Mini Shai-Hulud malware.  ( 56 min )

Adobe Acrobat Reader Escript.api Use-After-Free Remote Code Execution By Michele Campa Overview In this blog post we take a look at a use-after-free vulnerability found in Adobe Acrobat Reader’s Escript.api module in February 2025. This issue was patched on April 2026 and likely assigned CVE-2026-34621,  CVE-2026-34626 , or CVE-2026-34622. Disclaimer: Every offset and function name referenced in this blog ... Read more Adobe Acrobat Reader Escript.api Use-After-Free Remote Code Execution The post Adobe Acrobat Reader Escript.api Use-After-Free Remote Code Execution appeared first on Exodus Intelligence.  ( 16 min )

Here is an overview of content I published in May: Blog posts: Update: search-for-compression.py Version 0.0.7 SANS ISC Diary entries: Wireshark 4.6.5 Released YARA-X 1.16.0 Release Wireshark 4.6.6 Released Microsoft Access VBA YARA-X 1.17.0 Release  ( 11 min )

Visual Studio Code 1.123 introduces significant updates to its integrated AI capabilities and underlying platform architecture. The editor now runs on Electron 42, which incorporates Chromium 148 and Node.js 22.x to improve performance and security. For Windows administrators, a critical fix addresses a bug where CLI flags for folder and file URIs would fail silently if not positioned as the final argument. Source  ( 16 min )

Microsoft has overhauled the Copilot interface within the Microsoft 365 suite to provide a more streamlined and consistent user experience. The update transforms the standard text entry box into a task-aware workspace designed to handle complex prompts and large amounts of pasted content. These changes aim to reduce visual clutter by revealing advanced AI capabilities only when they are relevant to the user's current context. Source  ( 16 min )

Visual Studio Code 1.122 introduces the ability to use artificial intelligence features without requiring a GitHub account sign-in. This update expands "Bring Your Own Key" support, allowing users to connect their own AI model providers directly within the editor. By removing mandatory authentication, the platform now supports fully offline AI workflows using local tools like Ollama. Source  ( 16 min )

OpenAI has released a readability update for GPT-5.5 Instant to improve the natural flow and structure of generated responses. The model now prioritizes better text organization and reduces the frequency of long, repetitive bulleted lists. These changes aim to make the output more accessible and easier for users to digest during standard interactions. Source  ( 16 min )

OpenAI has updated its Codex application to allow autonomous interaction with the Windows 11 operating system. This "Computer Use" feature enables the AI to navigate the interface, manage files, and execute tasks without direct human intervention. Users can trigger these capabilities within the settings and use specific commands to target individual programs like Microsoft Paint. Source  ( 16 min )

Microsoft has released a fix for a persistent bug that prevented the installation of the May 2024 Patch Tuesday update, KB5089549. This issue primarily affected Windows 11 versions 24H2 and 25H2, causing systems to trigger a rollback during the reboot phase. The failure was characterized by error code 0x800f0922 and typically occurred when the installation process reached approximately 35 percent completion. Source  ( 16 min )

Microsoft Entra ID, the identity management system formerly known as Azure Active Directory, is implementing stricter security for its Self-Service Password Reset (SSPR) portal. Currently, users can reset passwords using contact details like phone numbers or alternate emails stored in their profiles without formal verification. Starting in September 2026, the system will only permit authentication methods that have been officially registered and verified by the user. Source  ( 16 min )

Nvidia is expected to unveil the N1X, its first consumer-focused CPU in over a decade, during a joint presentation with Microsoft at Computex 2026. This new system-on-chip utilizes the Arm architecture, which focuses on high power efficiency and reduced heat generation compared to traditional x86 processors. The collaboration aims to establish a new era for Windows 11 AI PCs by providing hardware capable of full Copilot+ acceleration and advanced local processing. Source  ( 16 min )

The Dutch National Police and the National Cyber Security Center have successfully dismantled a massive botnet consisting of more than 17 million compromised devices. This operation involved the seizure of several servers from a local hosting provider after a security researcher alerted authorities to the infrastructure. The network was managed through approximately 200 command-and-control servers located within the Netherlands to facilitate various cybercriminal activities. Source  ( 15 min )

The rapid escalation of artificial intelligence costs has forced corporations into an unprecedented dilemma where technology spend now rivals the expense of human employees. Enterprise leaders are increasingly making direct trade-offs between hiring new talent and funding the vast amounts of compute and tokens required to run frontier models. As annual AI budgets are often exhausted within months, the industry is shifting toward model-agnostic strategies to optimize unit economics and prevent unsustainable financial outflows. Source  ( 16 min )

New research reveals that leading AI search agents often suffer from intrinsic knowledge dependence, a phenomenon where models rely on pre-trained memory instead of active web browsing. While these agents achieve high scores on traditional benchmarks, their performance collapses when tasked with finding obscure information from the last 90 days. This suggests that current leaderboards may be measuring a model's internal database rather than its actual ability to navigate and synthesize the live web. Source  ( 16 min )

Our annual report on all the Mac malware of the year - including samples for download, infection vectors, persistence mechanisms, payloads and more!  ( 41 min )

NSLog(@"Hello World"); objective-see.org is alive!  ( 42 min )

Announcing the release of DHS; a tool to help detect (dylib) hijackers  ( 42 min )

Exploiting RootPipe on OS X 10.10.3  ( 42 min )

Details on bypassing Apple's original rootpipe patch  ( 42 min )

A deeper dive into 'MacInstaller' and the adware it installs  ( 42 min )

How to build HackingTeam's OS X implant in Xcode  ( 42 min )

How reversing Apple's 'RootPipe' patch provided the means to secure TaskExplorer's XPC service  ( 42 min )

How to remotely kernel-debug a OS X 10.11 VM  ( 42 min )

Why BlockBlock needs a kext (hint: process monitoring), and how the kext was created  ( 42 min )

Process monitoring via the KAuth Subsystem (and some limitations)  ( 5 min )

Let's look at all the mac malware from 2017, for each - discussing their infection vector, persistence mechanism, features & goals.  ( 40 min )

Our annual report on all the Mac malware of the year - including samples for download, infection vectors, persistence mechanisms, payloads and more!  ( 26 min )

Learn how to leverage Apple's new Endpoint Security Framework to create a comprehensive (user-mode) Process Monitor for macOS 10.15!  ( 14 min )

Getting process creation notifcations from kernel-mode to user-mode, via the undocumented kev_msg_post function  ( 8 min )

HackingTeam using native OS X crypto to protect malware -neat! New blog w/ sample + decryptions/dumpings/detections  ( 5 min )

Dissecting string obfuscations, junk code insertions, and anti-debugging logic of InstallCore  ( 5 min )

In Objective-See's first guest blog post, Amit Serper presents his detailed analysis of OSX/Pirrit  ( 10 min )

By monitoring file I/O events and detecting the rapid creation of encrypted files by untrusted processes, can ransomware be generically detected?  ( 26 min )

How to verify that an application came from the official Mac App Store, via receipt validation  ( 7 min )

Learn how a Finder Sync can 'extend' Finder.app and how this could be abused for persistence  ( 5 min )

The 'Mac File Opener' adware is fairly normal, except for it how it persists via registered document handlers  ( 9 min )

Does Shazam's Mac App keep recording even when you turn the app off? ...yes :/  ( 8 min )

Apple's App Translocation broke several of my tools, but we can locally undo it to restore broken functionality!  ( 11 min )

Read how an attacker can bypass Apple's SIP, via the local OS upgrade process  ( 11 min )

Let's analyse the malware that appeared in 2016, discussing the infection vector, persistence mechanism, feature, and disinfection for each.  ( 12 min )

A Word document targets Mac users with malicious macros and an open-source payload.  ( 4 min )

Today is our 2nd birthday! Let's look at our past, present, and future.  ( 2 min )

Reverse-engineering a 'Russian' implant reveals HackingTeam's code!?  ( 9 min )

Apple's 'fix' for a macOS kernel panic, fixes nothing and worse, introduces a new bug.  ( 13 min )

The macOS kernel had an (intentional?) off-by-one bug that could trigger a kernel panic.  ( 13 min )

Analyzing code within the macOS kernel audit subsystem uncovered an exploitable heap overflow.  ( 5 min )

The website of a popular application was hacked, and the application trojaned with a new variant of osx/proton.  ( 3 min )

Looks like somebody on the 'dark web' is offering 'Ransomware as a Service'...that's designed to infect Macs!  ( 6 min )

Analysis of OSX/Proton.B reveals some interesting tricks plus a command file that can be decrypted to reveal the malware's capabilities  ( 10 min )

A new 'security' feature in macOS 10.13, is trivial to bypass.  ( 5 min )

Some undetected adware named "Mughthesec" is infecting Macs...let's check it out!  ( 5 min )

High Sierra suffered from a nasty bug (CVE-2017-7149) that afforded local attackers access to the contents of encrypted APFS volumes.  ( 12 min )

Yet another a massive security flaw affects the latest version of macOS (High Sierra), allowing anybody to log into the root account with a blank, or password, of their choosing!  ( 7 min )

Here, we reverse, then 'extend' a popular macOS anti-virus engine. With the creation of a new anti-virus signature, classified documents will be automatically detected!  ( 14 min )

OSX/MaMi (the first Mac malware of 2018) hijacks infected users' DNS settings and installs a malicious certificate into the System keychain, in order to give remote attackers 'access' to all network traffic  ( 9 min )

On my flight to ShmooCon, I managed to panic my fully-patched MacBook. Here we analyze the kernel panic report, finding that Apple's AMDRadeonX4150 kext is responsible for the crash.  ( 7 min )

The EFF/Lookout discovered a cross-platform implant, named CrossRat with ties to nationstate operators. Here, we tear it apart; analyzing its persistence mechanisms, features, and network communications.  ( 13 min )

Recently, the popular MacUpdate website was subverted to distribute a new macOS cryptominer; OSX/CreativeUpdater.  ( 5 min )

Turns out the innocuously named "Calendar 2" app, found on the official Mac App Store, was surreptitiously turning Mac into cryptocurrency miners!  ( 4 min )

I uncovered a new cross-platform backdoor that provides remote attackers persistent access to infected systems  ( 12 min )

Turns out that writing security tools is a great way to inadvertently uncover bugs in macOS. How about a crash in Apple's 'Security' framework ... that can't be good!?  ( 7 min )

In this guest blog post my friend Mikhail Sosonkin reverses Apple's screencapture utility, discusses Mac malware that captures desktop images, and suggests methods for screen-capture detection!  ( 11 min )

Did you know on macOS, notifications are stored in a unencrypted database? Which means that even 'disappearing' messages from apps such as Signal - may not really disappear. Yikes!  ( 4 min )

In macOS Mojave apps, to have to obtain user permission before using the Mac camera & microphone. We'll illustrate how this is trivial to bypass (at least in the current beta).  ( 4 min )

A new Mac malware targets the cryptocurrency community. In this post, we dive into the malware and illustrate how Objective-See's tools can generically thwart this new threat at every step of the way.

Are full paths and preview thumbnails for files even on encrypted containers and removable usb devices really persistently stored? ...yes :( Apple's 'QuickLook' cache is to blame.  ( 7 min )

Apple recently updated the way login items are stored by the OS. In this post, we'll illustrate how to parse the (new) login item files to detect persistence

In this guest blog post @CodeColorist writes about a neat macOS vulnerability. Ironically, by abusing security mechanisms such as sandboxing, macOS can be coerced to load an untrusted library, into a SIP-entitled process!

Imagine you've gained remote code execution on a Mac via a malicious Word document. Turns out, you're still stuck in a sandbox. However, via a faulty regex, you can escape and persist!  ( 5 min )

If you can programmatically generate synthetic mouse clicks, you can break macOS! Approving kernel extensions, dismissing privacy alerts, and much more more...  ( 8 min )

Apple wrote code to appease the Chinese government ...it was buggy. In certain configurations, iOS devices were vulnerable a "emoji-related" flaw that could be triggered remotely!

The WINDSHIFT APT group is successfully infecting Macs with a novel infection mechanism. By abusing custom URL scheme handlers and minimal user interaction, Macs can be remotely compromised!  ( 8 min )

The macOS sandbox is seeks to prevent malicious applications from surreptitiously spy on unsuspecting users. Turns out, it's trivial to sidestep some of these protections, resulting in significant privacy implications!  ( 7 min )

A massively popular app from the official Mac App Store, surreptitiously steals your browsing history! By fully reversing the application, we can fully expose its functionality and rather shady capabilities.  ( 13 min )

A malicious Word document targeting macOS users, was recently uncovered. Let's extract the embedded macros, decode an embedded downloader, and retrieve the 2nd-stage payload!  ( 4 min )

The APT group WindShift has been targeting Middle Eastern governments with Mac implants. Let's analyze their 1st-stage macOS implant: OSX.WindTail!  ( 7 min )

The APT group WindShift has been targeting Middle Eastern governments with Mac implants. Let's (continue to) analyze their 1st-stage macOS implant: OSX.WindTail!  ( 8 min )

A core Mojave utility is rather disastrously broken - causing a full-system lockup. Let's find out why!  ( 6 min )

Let's tear apart a persistent piece of adware, decompiling, decoding, and decompressing it's code to uncover its methods and capabilities.  ( 8 min )

@CodeColorist continues writing about bugs, such as CVE-2019-8521 and CVE-2019-8565 that provide a mechanism to elevate privileges to root on macOS.  ( 4 min )

In part one of a guest blog post, @CodeColorist writes about several neat macOS vulnerabilities.  ( 4 min )

After the success of #OBTS v1.0, we decided to go international and plan #OBTS v2.0 in Europe! In this blog post, we re-live the highlights (from Monaco!) of "Objective by the Sea" v2.0.  ( 3 min )

Recently, an attacker targeted (Mac) users via a Firefox 0day. In this first post, we triage and identify the malware (OSX.NetWire.A) utilized in this attack, identifying its methods of persistence, and more!  ( 6 min )

Recently, an attacker targeted (Mac) users via a Firefox 0day. In this third post, we analyze a second backdoor used in the attack, detailing its persistence, capabilities, and ultimate identify it a new variant of the cross-platform Mokes malware!  ( 6 min )

Recently, an attacker targeted (Mac) users via a Firefox 0day. In this second post, we fully reverse OSX.NetWire.A, revealing (for the first time!), its inner workings and complex capabilities.  ( 12 min )

In this guest blog post, "Objective by the Sea" speaker, Csaba Fitzl writes about an interesting way to get root via Apps from the official Mac App Store!  ( 17 min )

A new macOS backdoor written by the infamous Lazarus APT group needs analyzing. Here, we examine it's infection vector, method of persistence, capabilities, and more!  ( 10 min )

Learn how to leverage Apple's new Endpoint Security Framework to create a comprehensive (user-mode) File Monitor for macOS 10.15!  ( 10 min )

The rather infamous APT group, "Lazarus", continues to evolve their macOS capabilities. Today, we tear apart their latest 1st-stage implant that supports remote download & in-memory execution of secondary payloads!  ( 8 min )

A 0day logic flaw in Microsoft Excel leads to 'remote' code execution on macOS, via malicious macros.  ( 6 min )

A massively popular iOS application turns out to be a government spy tool! Here, we analyze the app; decrypting its binary and studying its network traffic.  ( 12 min )

CVE-2017-7170 was a local priv-esc vulnerability that affected OSX/macOS for over a decade! Here (for the first time!), we dive into the technical details of finding the bug, the core flaw, and exploitation.  ( 10 min )

The Lazarus group's latest implant/loader supports in-memory loading of 2nd-stage payloads. In this post we describe exactly how to repurposing this 1st-stage loader to execute *our* custom 'fileless' payloads!  ( 14 min )

Today we uncover two (local) security flaws in Zoom's latest macOS client. First, a privilege escalation vulnerability, and second, a method to surreptitiously access a user's webcam and microphone (via Zoom).  ( 12 min )

A sophisticated Lazarus Group implant has arrived on macOS. In this post, we deconstruct the Mac variant of a OSX.Dacls, detailing its install logic, persistence, and capabilities.  ( 11 min )

Tool annotations in the Model Context Protocol (MCP) provide metadata hints that describe how tools behave, enabling AI clients to make informed decisions about safety, permissions, and user experience. Annotations also help AI systems understand whether a tool modifies data, operates destructively, or interacts with external systems. In the lethal trifecta problem—where agents have access to private data, untrusted content, and external connectivity-MCP annotations help improve the security of MCP servers. Source  ( 25 min )

Does your Microsoft stack cover every device in your environment? Jamf's seamless integrations can make it happen.  ( 15 min )

Failed deployments devour IT’s time and dominates productivity. Discover why the cycle keeps repeating – and learn how to break it.  ( 7 min )

A CVSS 10.0 path traversal in UniFi Network Application lets unauthenticated attackers read controller backups, extract credentials, and take over every managed device on the network. Bishop Fox breaks down the attack paths, the preconditions, and a safe detection tool to check your exposure.  ( 20 min )

Discovery During a recent network security assessment, we were working on an environment that was well-hardened – Patching was current, password policies were strong, and network segmentation was in place. So, as part of our enumeration of all network assets, we started looking for default credentials and this led us to multiple Canon enterprise printers […] The post When Encryption Isn’t Really Encryption appeared first on Praetorian.  ( 15 min )

In previous blog posts we’ve talked about getting nerd sniped. Today we’re going to talk about a kind of nerd sniping that any offensive security tool creator is familiar with; when your tool gets signatured. This normally kicks off a frustrating spiral of back and forth changes between the tool author and security vendors until […] The post Adversarial Oracles: LLM-Guided EDR Signature Reduction appeared first on Praetorian.  ( 19 min )

GitHub Copilot app is a desktop application for agentic development that provides a centralized workspace to manage AI agents across parallel workflows, integrate with GitHub issues and pull requests, and handle the entire development lifecycle without switching between terminals, IDEs, and browser tabs. The app is built on top of GitHub Copilot CLI and integrates directly with GitHub repositories. It supports Windows, macOS, and Linux, and requires a paid GitHub Copilot subscription. This article explains what the app does, how to access it, and its current limitations. Source  ( 23 min )

Key Takeaways AI generates findings at scale, but scale without trust creates risk. The real security challenge isn’t discovery—it’s knowing which findings are real, exploitable, and worth acting on before automated systems take action. False positives become operationally dangerous in AI-driven environments. Model hallucination, single-tool reliance, and misinterpreted context can cause AI to fabricate vulnerabilities […] The post AI Can’t Fix What It Can’t Trust: Why Continuous Security Validation Matters appeared first on Synack.  ( 14 min )

TL;DR: Building a security program and exercising it are not the same investment. Most organizations prioritize the first and defer the second. This post explains what structured practice requires and why it belongs in every mature security program.  Most security teams have runbooks, escalation procedures, and documented response processes. Far fewer have practiced executing those processes under realistic […] The post The Case for Practicing Response Before You Need It appeared first on SpecterOps.  ( 13 min )

TL;DR: Phishing sandboxes are a pain. Cloudflare Turnstile can be used as an effective solution to conceal your phishing pages. Intro Recently, I was on a red team engagement that involved email phishing. I thought to myself, “No problem! I have done phishing before relatively successfully.” Instead, I discovered that the landscape drastically changed from […] The post Don’t Jump the Turnstile: Lessons from the Field appeared first on SpecterOps.  ( 18 min )

Discussion of PQC relevant statistics that we see across our customers and other data sources.  ( 61 min )

Defenders don’t need to detect every adversary action to prevent a threat. Here’s a more realistic, optimized approach to testing.  ( 30 min )

﷽  ( 10 min )

Microsoft released a new TeamViewer remote assistance connector for Intune in April 2026, replacing the older integration available since 2016. The new connector requires TeamViewer Tensor or a compatible enterprise license with a Microsoft add-on, and it requires the remote device to be actively managed by TeamViewer, not just enrolled in Intune. This article explains what changed, what you need to configure the integration, and what the new connector cannot do. Source  ( 25 min )

Phishing is about creativity. The less likely your target is to think about a link being potentially malicious, the more likely you are to have success. Read how our creative Social Engineering experts ruined free cookies in the break room. The post Phishing with Misfortune Cookies  appeared first on NetSPI.  ( 13 min )

Wiz Sensor Forensics is now generally available - automatically capturing forensic artifacts at the moment of detection and using AI to accelerate investigation for SOC and IR teams.  ( 57 min )

Wiz CIRT and Wiz Research detail JINX-0164, a threat actor using LinkedIn social engineering, custom macOS malware, and CI/CD hijacking to target cryptocurrency organizations.  ( 74 min )

How Wiz helps organizations adopt an AI Operating Model for AI Threat Readiness  ( 60 min )

TL;DR : Splunk is a daunting SIEM to learn, but this learning curve can be flattened by learning to use the basic building blocks which make up most Splunk searches. A Detection Engineer’s Guide When I first got my start in cybersecurity, Splunk was one of the most daunting platforms to use. The SIEM had […] The post Spelunking through Splunk appeared first on SpecterOps.  ( 21 min )

Read our primer on how to detect and respond to an autonomous agent escalating privileges and persisting in your Entra ID tenant  ( 40 min )

Microsoft SQL Server 2025 and Azure SQL Database introduce comprehensive artificial intelligence capabilities that transform how you build intelligent applications. These include native vector database support, external AI model integration, and agentic AI features that let you implement advanced AI scenarios directly within your database environment. Source  ( 26 min )

Doyensec performed a side-by-side comparison of two leading AI-powered penetration testing platforms: Aikido’s Attack AI Pentest and XBOW’s Lightspeed in order to evaluate their abilities to properly identify vulnerabilities in modern web applications. This included manually validating all findings and classifying them as either true positives or false positives. Additionally, we looked at their overall testing process, including the configuration, impact on tested applications, quality and content of the reports, cost, and speed. As a leading boutique application security consultancy, we were also curious about how the adoption of AI will impact the future of testing. To understand the current maturity levels of these AI platforms, it was necessary for us to put some vendors’ claims to the test. If you’re interested in the current state of AI-powered pentesting, we encourage you to give it a read: Comparing AI Application Security Testing Platforms: Aikido vs. XBOW (PDF, 5 MB)  ( 2 min )

Jamf School makes it simple for K-12 schools to deploy and manage Mac alongside iPad — delivering a secure, seamless, education-focused experience that empowers students and reduces IT complexity.  ( 6 min )

TL;DR  Background  I recently wrote a blog post on the Shelly Gen 4 open access point issue. While I was investigating that, I started looking more broadly at the Bluetooth configuration across Shelly’s modern device range and found another issue, this time with the Wall Display.  The modern Shelly devices now include Bluetooth and by default […] The post Shelly Wall Display exposed RPC over Bluetooth  appeared first on Pen Test Partners.  ( 6 min )

ClearFake is in command and ACR Stealer and GraphRunner debut in this month’s edition of Intelligence Insights  ( 29 min )

Sparkplug B is the dominant protocol in ICS and SCADA environments, but no public security fuzzer existed for it until now. Bishop Fox used AI-assisted development to build one from scratch, covering all 9 message types, 19 data types, and 87+ field paths from the full specification.  ( 13 min )

Insights from real-world environments into how code, developer tooling, automation, and AI are reshaping application security.  ( 53 min )

﷽  ( 9 min )

Tycoon 2FA bypasses MFA on Entra ID and Google Workspace. We map telemetry fingerprints across both platforms, ship detection rules for both tiers, and contain incidents in under 10 seconds with Elastic Workflows.  ( 38 min )

Proxmox VE 9.2, released on May 21, 2026, adds cluster load balancing, software-defined networking changes, custom CPU model management, High Availability maintenance controls, and updated virtualization components. Proxmox VE is a virtualization platform for running virtual machines and Linux containers on bare-metal servers. The main technical changes are relevant if you operate clustered hosts, use Proxmox High Availability, or manage routed virtual networks. This is not a redesign release, but it changes several operational areas that administrators should review before upgrading. Source

VMware Workstation Pro 26H1 is a desktop virtualization update for hosts running Windows and Linux. For Windows administrators, the main changes are the 64-bit Windows application, updated virtual machine metadata, support for remote connections to ARM-based ESXi hosts, and new supported Linux host and guest systems. Broadcom released Workstation Pro 26H1 on May 14, 2026, with build 25388281, and updated the release notes on May 15, 2026. Source

﷽  ( 8 min )

After our last episode on Multi-SSO Cognito User Pools, we are back with another issue. This time, we are looking at one of those AWS components that is everywhere and rarely questioned deeply enough: the Elastic Load Balancer. Tidbit No. 5 - Navigating Lax Load Balancers What is AWS ELB? AWS Elastic Load Balancing (ELB) distributes traffic to backend services and serves as the entry point between the Internet and your applications. It supports Layer 7 routing (Application Load Balancer - ALB) and Layer 4 routing (Network Load Balancer - NLB). It decides where traffic goes and under which conditions. ELB is commonly found fronting multiple applications, environments, and trust zones across the same infrastructure. Why It Matters ELB is often the first public entry point before applicati…  ( 7 min )

This is a small bug fix for search-for-compression.py, and I’m also taking it out of the beta repository and putting it into the DidierStevensSuite repository. search-for-compression_V0_0_7.zip (http)MD5: DD113FF41851A562D271804E4558EA08SHA256: 6E663316F774BA5B373704E7FE41B8266F5D1ADC618327F9F2C4C4C830A1B3DE  ( 11 min )

Microsoft announced on May 8, 2026, that it will retire direct certificate-based authentication (CBA) for Exchange ActiveSync (EAS) by the end of 2026. If your organization uses certificates to authenticate mobile devices against Exchange Online, you must migrate to a new method before the deadline, or your users' mobile email will stop working. This article explains what the change means, who is affected, and what steps you need to take. Source

Read a deep technical exploration of ARM's Memory Tagging Extension, Apple's Memory Integrity Enforcement (MIE), and how hardware-assisted memory safety is reshaping kernel security and vulnerability research across modern ARM architectures.  ( 16 min )

The RAG bot, with checkpoints Let’s circle back to the team from our introduction. With the three checkpoints in place, the same attack would have been intercepted three different ways: Three layers, three different ways to catch the same attack. That is what defense in depth means in this stack: not a single perfect filter, … Continue reading Securing AI systems without overconfidence or fear – Part 2: Attack surfaces and the checkpoint flow →  ( 16 min )

CVE-2026-0265 lets unauthenticated attackers forge a JWT and log in as any trusted user on CAS-enabled PAN-OS deployments. Bishop Fox built a detection tool that returns a definitive verdict from a single anonymous request, and breaks down exactly how the bug works and what to do about it.  ( 12 min )

A sanitization bypass in Strapi 4.0.0 through 5.36.1 lets unauthenticated attackers extract an admin's password reset token character by character and take over the account. With over 20,000 internet-facing hosts exposed, Bishop Fox breaks down how the exploit works and how to remediate it.  ( 17 min )

In March 2026, attackers exploited a pull_request_target misconfiguration in the aquasecurity/trivy-action GitHub Action to exfiltrate organization and repository secrets, then used those credentials to backdoor LiteLLM on PyPI (see Trivy’s post-mortem for the full timeline). zizmor is a static analyzer that GitHub Actions users run to catch exactly these misconfigurations before they ship. When GitHub Actions added support for YAML anchors in September 2025, a small but high-value slice of the ecosystem started writing workflows that zizmor could only analyze on a best-effort basis. Over the past three months, Trail of Bits collaborated with the zizmor maintainers to bring zizmor’s anchor support up to full coverage. First, we fixed parsing bugs that caused crashes, produced wrong-locatio…  ( 4 min )

Elastic Security Labs presents a detailed reverse-engineering analysis of PHANTOMPULSE, the long-lived RAT delivered to crypto-sector victims through the REF6598 intrusion set.  ( 53 min )

Microsoft is shutting down Exchange Web Services (EWS) — the nearly 20-year-old API that Exchange uses for hybrid coexistence — in Exchange Online in two phases: a soft block on October 1, 2026, and a permanent hard shutdown on April 1, 2027. If you run Exchange in hybrid mode, meaning some mailboxes are on-premises and some are in Microsoft 365, this requires a two-step migration. The first step should already be complete; the second step must be finished before October 2026 and requires Exchange Server Subscription Edition (SE). Microsoft has confirmed there will be no exceptions past April 2027. Source

Microsoft now offers Windows Server 2025 hotpatching through Azure Arc at no additional charge for eligible Azure Arc-enabled servers. Hotpatching installs Windows security updates without restarting the server in most months, but it does not eliminate all reboots. You still need Azure Arc, the Azure Connected Machine agent, Virtualization-based Security, and a supported Windows Server 2025 edition. This article explains what those requirements mean, how to enable the feature, and where its limits are. Source

Key Takeaways Vulnerabilities Report Offers Key Industry Benchmarks How does your MTTR hold up against the industry average? And does your organization encounter more high/critical vulnerabilities than others in your industry? Those are just a few questions that our 2026 State of Vulnerabilities Report answers. The report analyzes more than 11,000 vulnerabilities surfaced through the […] The post The 2026 State of Vulnerabilities Report: Industry Insights appeared first on Synack.  ( 11 min )

Explore the intricacies of UEFI security with exploration into emulation, dynamic analysis, and the LogoFail vulnerability. Learn how subtle input manipulations can expose critical firmware weaknesses. The post Emulating & Exploiting UEFI: Unveiling Vulnerabilities in Firmware Security appeared first on NetSPI.  ( 29 min )

Security and compliance teams can now monitor Claude activity directly in Wiz, extending the workflows they already rely on to AI  ( 53 min )

TL;DR: TailscaleHound is an OpenGraph collector for BloodHound that maps Tailscale users, devices, groups, tags, ACLs, grants, SSH rules, routes, app connectors, services, keys, invites, webhooks, and hybrid Azure identity relationships. The result is a graph that helps answer practical questions like, “Which users can reach this device?”, “Who can use this exit node?”, “Which […] The post Introducing TailscaleHound: Mapping Tailscale Attack Paths in BloodHound appeared first on SpecterOps.  ( 16 min )

Background While discussing with eversinc33 about lifting BinaryShield to LLVM IR I decided it would be useful to write a basic lifter in Python that can lift x86_64 instructions to LLVM IR. He has since released his blog post: Writing a Naive LLVM-based Devirtualizer, which I highly recommend you check out! This post assumes familiarity with the basics of LLVM IR. You can find some references at the end of this post. Over the years I noticed that a lot of people get stuck exploring lifters, because existing tooling is too difficult to compile. In October 2025 I spent around a month redoing Remill’s build system (remill#723) and earlier this month I did the same for the Dna project (Dna#9). Last year I also started working on Python bindings for LLVM, which I wanted to use for a real project. You can find the lifter at LLVMParty/striga. The goal of this post is to lower the barrier of entry and let you experiment with lifting to LLVM IR. For inspiration you can look at the Static Devirtualization of Themida post that was just released by Back Engineering Labs, as well as the Pushan: Trace-Free Deobfuscation of Virtualization-Obfuscated Binaries paper by ASU researchers published in March. If you enjoy this article and would like to learn more, see my website for information about my in-person trainings. Lifting Lifting is the process of translating assembly instructions to some kind of intermediate representation (IR). The motivation is usually that directly analyzing and manipulating (x86) assembly instructions is complex and error prone. The lifter translates the underlying instruction semantics directly to an IR that is easier to reason about (and therefore to manipulate as well). A few popular IRs: SMT-LIB, used by Triton (symbolic execution) VEX, used by angr Miasm IR Sleigh, used by Ghidra, Remill and Icicle LLVM IR, used by Rellume, revng and Remill Microcode, used by IDA (proprietary) BNIL, used by Binary Ninja (proprietary) For this project I picked LLVM IR, because I am the most familiar with it and it has a well-established ecosystem. LLVM already has all of the common compiler optimizations and it is used and maintained by teams at large corporations. Architecture The architecture of the lifter is very much inspired by remill, but I simplified some things to make it easier to follow. In LLVM a register is actually an SSA value, which means we can only assign to it once. CPU registers are variables that can be assigned to multiple times. We model this by creating a State structure in memory that represents the x86 CPU state: struct State { uint64_t rax; uint64_t rbx; uint64_t rcx; uint64_t rdx; // ... GPRs uint8_t cf; uint8_t zf; uint8_t of; // ... Flags // ... XMM }; Instructions that read or write to RAX will load/store to State->rax. If we play our cards right, the optimizer will use the mem2reg pass to translate this into SSA form for us and enable further optimizations. An important difference to an actual CPU is that flags are modelled as independent 8-bit registers. This makes it easier to reason about compared to a packed bitfield. For instance, it helps the optimizer to perform dead store elimination and propagation. In addition to the State, we need an opaque memory pointer that helps us differentiate a load/store in the State from memory accesses by the x86 CPU. In short: the State pointer is used to model the CPU and the memory pointer is used to model the RAM. While lifting, the prototype of the lifted function is void lifted(State* state, void* memory). Later on we will perform brightening, to turn this into something we can recompile. Below is the LLVM IR for the instruction mov rax, rcx, with comments in pseudo-C: define internal void @lifted_0x140001000(ptr %state, ptr %memory) { initialize: ; uint64_t* rcx = &state->rcx; %rcx = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 2 ; uint64_t* rax = &state->rax; %rax = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 0 ; Jump to the first instruction br label %insn_0x140001000 insn_0x140001000: ; preds = %initialize ; uint64_t v0 = *rcx; %0 = load i64, ptr %rcx, align 4 ; *rax = v0; store i64 %0, ptr %rax, align 4 ; Jump to the next instruction br label %insn_0x140001003 insn_0x140001003: ; preds = %insn_0x140001000 ; Block terminator to keep the IR valid ret void } We start out with the initialize block, which is used to get pointers to the relevant State members. Then every instruction gets its own basic block named insn_<addr>. Every instruction is responsible for emitting an unconditional branch to its successors. The basic block for the successor is created with just a ret terminator, to keep the module verifier happy. To illustrate memory accesses, here is the LLVM IR for mov rax, qword [rbx+42]: define internal void @lifted_0x140001000(ptr %state, ptr %memory) { initialize: %rbx = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 1 %rax = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 0 br label %insn_0x140001000 insn_0x140001000: ; preds = %initialize ; uint64_t v0 = *rbx; %0 = load i64, ptr %rbx, align 4 ; uint64_t v1 = v0 + 42; %1 = add i64 %0, 42 ; uint8_t* v2 = &memory[v1]; %2 = getelementptr i8, ptr %memory, i64 %1 ; uint64_t v3 = *(uint64_t*)v2; %3 = load i64, ptr %2, align 1 ; *rax = v3; store i64 %3, ptr %rax, align 4 br label %insn_0x140001004 insn_0x140001004: ; preds = %insn_0x140001000 ret void } Here you can see the getelementptr i8, ptr %memory, i64 %1 instruction which uses memory as a base, signaling that this is a read from the x86 memory (we will clean this up later). The lifter itself is contained in a ~500 line Semantics class with these main functions (some are omitted for brevity): # src/striga/semantics.py class Semantics: def __init__(self, module: Module): ... # Lifting def begin(self, address: int) -> Function: ... def get_or_create_block(self, address: int) -> BasicBlock: ... def lift_bytes(self, address: int, code: bytes) -> list[Successor]: ... # Semantic helpers def reg_read(self, name: str) -> Value: ... def reg_write(self, name: str, value: Value): ... def mem_read(self, addr: Value, ty: Type) -> Value: ... def mem_write(self, addr: Value, value: Value): ... def op_mem(self, op: X86Op) -> Value: ... def op_read(self, index: int) -> Value: ... def op_write(self, index: int, value: Value): ... def flag_read(self, name: str) -> Value: ... def flag_write(self, name: str, value: Value): ... # State (simplified) module: Module function: Function ir: Builder insn: CsInsn The begin(address) function is used to create the lifted_<address> function in LLVM IR and create the initialize block with a branch to the first instruction: def begin(self, address: int) -> Function: name = f"lifted_{hex(address)}" fn = self.module.get_function(name) if fn is None: fn = self.module.add_function(name, self.lifted_ty) fn.param_attributes(0).add("noalias") fn.param_attributes(1).add("noalias") state, memory = fn.params memory.name = "memory" state.name = "state" self.function = fn self.reg_ptrs = {} self.insn_blocks = {} entry = fn.append_basic_block("initialize") assert fn.last_basic_block == entry with entry.create_builder() as ir: ir.br(self.get_or_create_block(address)) else: # Omitted for brevity return self.function To create the instruction block, get_or_create_block is used: def get_or_create_block(self, address: int) -> BasicBlock: block = self.insn_blocks.get(address) if block is None: block = self.function.append_basic_block(f"insn_{hex(address)}") with block.create_builder() as ir: ir.ret_void() self.insn_blocks[address] = block assert block.function == self.function return block As mentioned above, an empty block is not valid LLVM IR so we populate it with a ret instruction. When actually lifting into the basic block, that instruction will be replaced with the lifted code. To lift a single instruction we pass its address and bytes to lift_bytes, which is responsible for producing LLVM IR: def lift_bytes(self, address: int, code: bytes) -> list[Successor]: # Ensure we have a function to lift into if not hasattr(self, "function"): self.begin(address) insn = self.cs_disasm(address, code) if self.verbose: print(";", hex(insn.address), insn.mnemonic, insn.op_str) # Skip lifting if the block is already populated block = self.get_or_create_block(address) assert block.first_instruction if block.first_instruction.opcode == Opcode.Ret: block.first_instruction.erase_from_parent() else: return [] with block.create_builder() as ir: # State used by semantic handlers self.ir = ir self.insn = insn handler = _semantics.get(insn.mnemonic) if handler is None and insn.mnemonic.startswith("lock "): # LOCK preserves the single-threaded architectural result; the # lifter does not model inter-thread atomicity separately. handler = _semantics.get(insn.mnemonic.removeprefix("lock ")) if handler is None: raise NotImplementedError(insn.mnemonic) successors = handler(self) if successors is None: # Linear fallthrough - handler didn't emit a terminator. fallthrough = address + insn.size ir.br(self.get_or_create_block(fallthrough)) successors = [Successor(address, self.const64(fallthrough))] # Make sure the handler produced valid IR self.module.verify_or_raise() return successors The function first ensures an empty insn_<address> block by removing the temporary ret instruction. Then it creates an IR Builder and calls the handler responsible for producing IR for the instruction being lifted (more on that below). If the handler does not return successors, lift_bytes handles the common fallthrough case by creating a basic block for the next instruction. It is up to the caller to handle the list of Successor tuples: class Successor(NamedTuple): src: int dst: Value We use an LLVM Value for the branch destination, because it is not always concrete (for example jmp reg). The semantic handlers are registered globally: # src/striga/semantic.py SemanticFn: TypeAlias = Callable[["Semantics"], list[Successor] | None] _semantics: dict[str, SemanticFn] = {} def semantic(fn: SemanticFn): name = getattr(fn, "__name__") _semantics[name.removesuffix("_")] = fn return fn # src/striga/x86/data.py @semantic def mov(sem: Semantics): value = sem.op_read(1) sem.op_write(0, value) Every handler gets an instance of Semantics, to allow easy access to x86 constructs like operands, registers, flags and memory. For example, op_read is implemented as follows: def op_read(self, index: int) -> Value: op: X86Op = self.insn.operands[index] if op.type == CS_OP_REG: name = self.reg_name(op.reg) # pyright: ignore[reportAssignmentType] return self.reg_read(name) if op.type == CS_OP_IMM: return self.const_n(op.imm, op.size * 8) if op.type == CS_OP_MEM: addr = self.op_mem(op) return self.mem_read(addr, self.types.int_n(op.size * 8)) assert False For our example mov rax, rcx, the function will forward to reg_read: def reg_read(self, name: str) -> Value: if name in self.reg_types: load = self.ir.load(self.reg_types[name], self.reg_ptr(name)) load.metadata["tbaa"] = self.tbaa_tags[name] return load full_name, size, bit_offset = self.subregs[name] load = self.ir.load(self.reg_types[full_name], self.reg_ptr(full_name)) load.metadata["tbaa"] = self.tbaa_tags[full_name] if bit_offset: load = self.ir.lshr(load, self.const64(bit_offset)) return self.ir.trunc(load, self.types.int_n(size)) This function transparently handles accesses to sub registers like eax, ax, al and ah and it returns an LLVM Value containing the loaded register value. The last missing piece is the reg_ptr function, which is responsible for creating the getelementptr in the function entry: def reg_ptr(self, name: str) -> Value: reg_ptr = self.reg_ptrs.get(name) if reg_ptr is not None: return reg_ptr entry = self.function.entry_block state = self.function.get_param(0) with entry.create_builder() as ir: ir.position_before(entry.terminator) reg_ptr = ir.struct_gep(self.state_ty, state, self.reg_indices[name], name) self.reg_ptrs[name] = reg_ptr return reg_ptr To help the optimizer we add TBAA Metadata to the register load/store instructions. In this case we know that a register loads/stores never alias with each other. By telling the optimizer about this, it can perform more aggressive dead-store elimination when optimizing a sequence of lifted instructions. Semantics So far we discussed the architecture of the lifter, but we only discussed the mov instruction so far. Almost every other instruction has more complex behavior, especially around flag handling. For instance here are the implementations of and/or/xor: # src/striga/x86/bitwise.py def write_logical_flags(sem: Semantics, result: Value): false = sem.const_n(0, 1) sem.flag_write("cf", false) sem.flag_write("pf", sem.result_parity_even(result)) sem.flag_write_undef("af") sem.flag_write("zf", sem.result_is_zero(result)) sem.flag_write("sf", sem.result_sign_bit(result)) sem.flag_write("of", false) def logical_binop(sem: Semantics, opcode: Opcode): dst = sem.op_read(0) src = sem.resize_int(sem.op_read(1), dst.type) result = sem.ir.binop(opcode, dst, src) sem.op_write(0, result) write_logical_flags(sem, result) @semantic def and_(sem: Semantics): logical_binop(sem, Opcode.And) @semantic def or_(sem: Semantics): logical_binop(sem, Opcode.Or) @semantic def xor(sem: Semantics): logical_binop(sem, Opcode.Xor) For reference here is the lifted LLVM IR for xor rax, rbx and the Python code responsible for each part: insn_0x140001000: ; preds = %initialize ; dst = sem.reg_read(0) %0 = load i64, ptr %rax, align 4 ; src = sem.resize_int(sem.op_read(1), dst.type) %1 = load i64, ptr %rbx, align 4 ; result = sem.ir.binop(Opcode.Xor, dst, src) %2 = xor i64 %0, %1 ; sem.op_write(0, result) store i64 %2, ptr %rax, align 4 ; sem.flag_write("cf", false) store i8 0, ptr %cf, align 1 ; sem.flag_write("pf", sem.result_parity_even(result)) %3 = trunc i64 %2 to i8 %4 = lshr i8 %3, 4 %5 = xor i8 %3, %4 %6 = lshr i8 %5, 2 %7 = xor i8 %5, %6 %8 = lshr i8 %7, 1 %9 = xor i8 %7, %8 %10 = and i8 %9, 1 %11 = icmp eq i8 %10, 0 %12 = zext i1 %11 to i8 store i8 %12, ptr %pf, align 1 ; sem.flag_write_undef("af") %13 = call i1 @__striga_undef_af(i64 5368713216) %14 = zext i1 %13 to i8 store i8 %14, ptr %af, align 1 ; sem.flag_write("zf", sem.result_is_zero(result)) %15 = icmp eq i64 %2, 0 %16 = zext i1 %15 to i8 store i8 %16, ptr %zf, align 1 ; sem.flag_write("sf", sem.result_sign_bit(result)) %17 = lshr i64 %2, 63 %18 = trunc i64 %17 to i1 %19 = zext i1 %18 to i8 store i8 %19, ptr %sf, align 1 ; sem.flag_write("of", false) store i8 0, ptr %of, align 1 ; Semantics.lift_bytes br label %insn_0x140001003 If you pay close attention you see a call to __striga_undef_af, which is a custom intrinsic used to represent something that has no clear analog in LLVM IR. In this case the description of the xor instruction says: The OF and CF flags are cleared; the SF, ZF, and PF flags are set according to the result. The state of the AF flag is undefined. This means that Intel/AMD does not want to document exactly how the value of AF is computed in silicon. In practice this can vary between CPU models/generations and it can be used as an anti-emulation trick, but we will not go into detail in this post. We emit __striga_undef_af, to allow the user to handle this however they see fit. If you are interested there is remill#766 with a little discussion about how to model this correctly. Another class of instructions to highlight here is the various branch instructions: # src/striga/x86/control.py def conditional_jump(sem: Semantics, cond: Value): brtrue = sem.insn.operands[0].imm brfalse = sem.insn.address + sem.insn.size sem.ir.cond_br( cond, sem.get_or_create_block(brtrue), sem.get_or_create_block(brfalse), ) src = sem.insn.address return [ Successor(src, sem.const64(brtrue)), Successor(src, sem.const64(brfalse)), ] def jcc(sem: Semantics, cc: str): return conditional_jump(sem, cc_cond(sem, cc)) @semantic def je(sem: Semantics): return jcc(sem, "e") @semantic def jmp(sem: Semantics): dst = sem.op_read(0) if dst.is_constant: sem.ir.br(sem.get_or_create_block(dst.const_zext_value)) else: sem.ir.call(sem.jmp_handler, [dst]) sem.ir.ret_void() return [Successor(sem.insn.address, dst)] @semantic def call(sem: Semantics): dst = sem.op_read(0) fallthrough = sem.insn.address + sem.insn.size sem.push(sem.const64(fallthrough)) sem.ir.call(sem.call_handler, [dst]) sem.ir.br(sem.get_or_create_block(fallthrough)) return [Successor(sem.insn.address, sem.const64(fallthrough))] @semantic def ret(sem: Semantics): dst = sem.pop(sem.i64) if sem.insn.operands: rsp = sem.reg_read("rsp") sem.reg_write("rsp", sem.ir.add(rsp, sem.const64(sem.insn.operands[0].imm))) sem.ir.call(sem.ret_handler, [dst]) sem.ir.ret_void() return [Successor(sem.insn.address, dst)] LLVM IR for je imm: insn_0x140001000: ; preds = %initialize %0 = load i8, ptr %zf, align 1 %1 = icmp ne i8 %0, 0 br i1 %1, label %insn_0x140001014, label %insn_0x140001002 insn_0x140001014: ; preds = %insn_0x140001000 ret void insn_0x140001002: ; preds = %insn_0x140001000 ret void } Note that the semantic handler for jcc is responsible for creating both the destination blocks as well as the br with the appropriate condition based on the flag(s). LLVM IR for jmp rbx: insn_0x140001000: ; preds = %initialize %0 = load i64, ptr %rbx, align 4 call void @__striga_jmp(i64 %0) ret void LLVM IR for call imm: insn_0x140001000: ; preds = %initialize %0 = load i64, ptr %rsp, align 4 %1 = sub i64 %0, 8 store i64 %1, ptr %rsp, align 4 %2 = getelementptr i8, ptr %memory, i64 %1 store i64 5368713221, ptr %2, align 1 call void @__striga_call(i64 5369761797) br label %insn_0x140001005 LLVM IR for ret: insn_0x140001000: ; preds = %initialize %0 = load i64, ptr %rsp, align 4 %1 = getelementptr i8, ptr %memory, i64 %0 %2 = load i64, ptr %1, align 1 %3 = add i64 %0, 8 store i64 %3, ptr %rsp, align 4 call void @__striga_ret(i64 %2) ret void } As you can see, we use the following intrinsics: __striga_jmp: indirect jump __striga_call: call instruction __striga_ret: ret instruction These are also used to give the user flexibility in how they want to handle these instructions. Control flow Because of the design choice where every instruction is a basic block, it becomes fairly straightforward to recover the control flow of a basic function: def lift(module: Module, container: Container, start: int, *, verbose=True): sem = Semantics(module, verbose=verbose) lifted_fn = sem.begin(start) queue: Queue[Successor] = Queue() queue.put(Successor(0, sem.const64(start))) # Keep destinations as LLVM Values instead of splitting constants into ints. # This keeps the worklist uniform and matches later slicing/data-flow uses. visited: set[Value] = set() while not queue.empty(): src, dst = queue.get() if not dst.is_constant: if sem.verbose: print(f"; non-constant branch destination: {hex(src)} -> {dst}") continue if dst in visited: continue visited.add(dst) va = dst.const_zext_value code = container.get_data(va, 15) successors = sem.lift_bytes(va, code) for successor in successors: if successor.dst in visited: continue queue.put(successor) sem.module.verify_or_raise() return lifted_fn This is a simple Breadth-first search over the control flow graph and it allows recovering functions without indirect branches. Note that we do not have to do anything special to handle back edges (loops) or block splitting. The lifted code is modeled with an LLVM basic block per instruction, so we can connect instructions arbitrarily. Below is a function with some simple control flow (if/else/loop): test_cfg: cmp rax, 0 je .else_block .if_true: add rax, 1 jmp .merge .else_block: add rax, 2 .merge: sub rax, 1 jne .merge .exit: ret The graph of the disassembly looks like this: The LLVM IR looks like this: define internal void @lifted_0x140001000(ptr %state, ptr %memory) { initialize: %rax = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 0 %zf = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 51 %rsp = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 6 br label %insn_0x140001000 insn_0x140001000: ; preds = %initialize ; cmp rax, 0 %0 = load i64, ptr %rax, align 4 %1 = sub i64 %0, 0 %19 = icmp eq i64 %1, 0 %20 = zext i1 %19 to i8 store i8 %20, ptr %zf, align 1 br label %insn_0x140001004 insn_0x140001004: ; preds = %insn_0x140001000 ; je 0x14000100c %30 = load i8, ptr %zf, align 1 %31 = icmp ne i8 %30, 0 br i1 %31, label %insn_0x14000100c, label %insn_0x140001006 insn_0x14000100c: ; preds = %insn_0x140001004 ; add rax, 2 %32 = load i64, ptr %rax, align 4 %33 = add i64 %32, 2 store i64 %33, ptr %rax, align 4 br label %insn_0x140001010 insn_0x140001006: ; preds = %insn_0x140001004 ; add rax, 1 %62 = load i64, ptr %rax, align 4 %63 = add i64 %62, 1 store i64 %63, ptr %rax, align 4 br label %insn_0x14000100a insn_0x140001010: ; preds = %insn_0x140001014, %insn_0x14000100a, %insn_0x14000100c ; sub rax, 1 %92 = load i64, ptr %rax, align 4 %93 = sub i64 %92, 1 store i64 %93, ptr %rax, align 4 %111 = icmp eq i64 %93, 0 %112 = zext i1 %111 to i8 store i8 %112, ptr %zf, align 1 br label %insn_0x140001014 insn_0x14000100a: ; preds = %insn_0x140001006 ; jmp 0x140001010 br label %insn_0x140001010 insn_0x140001014: ; preds = %insn_0x140001010 ; jne 0x140001010 %122 = load i8, ptr %zf, align 1 %123 = icmp ne i8 %122, 0 %124 = xor i1 %123, true br i1 %124, label %insn_0x140001010, label %insn_0x140001016 insn_0x140001016: ; preds = %insn_0x140001014 ; ret %125 = load i64, ptr %rsp, align 4 %126 = getelementptr i8, ptr %memory, i64 %125 %127 = load i64, ptr %126, align 1 %128 = add i64 %125, 8 store i64 %128, ptr %rsp, align 4 call void @__striga_ret(i64 %127) ret void } For clarity, some flag computations were omitted from this IR dump. Brightening Brightening was a term coined in 2019 by Peter Garba and Matteo Favaro in the SATURN paper: Brightening [COMP.] verb – Reshaping code to make it more readable and understandable for humans Concretely it means to transform the LLVM IR from the lifted shape (pseudo C): /* Lifted instructions: add rdi, rsi mov rax, rdi ret */ void lifted(State* state, void* memory) { state.rdi += state.rsi; state.rax = state.rdi; __striga_ret(...); } Back to a regular function for the lifted platform’s calling convention, such as: // Linux calling convention: https://wiki.osdev.org/System_V_ABI#x86-64 uint64_t /* rax */ brightened(uint64_t /* rdi */ x, uint64_t /* rsi */ y) { return x + y; } The brightened function sets up the State on the stack and assigns the arguments to the registers appropriate for the calling convention of our target platform. The result register is returned from the function. Conceptually this is not very difficult, but it requires a bit of mental gymnastics to wrap your head around the trick: // Symbolic variable for memory uint8_t RAM[0]; void lifted(State* state, void* memory) { ... } uint64_t brightened(uint64_t x, uint64_t y) { State state; state.rdi = x; state.rsi = y; lifted(&state, RAM); return state.rax; } After an inlining pass it would look something like this: uint64_t brightened(uint64_t x, uint64_t y) { State state; state.rdi = x; state.rsi = y; state.rdi += state.rsi; state.rax = state.rdi; __striga_ret(...); return state.rax; } We can get rid of the __striga_ret intrinsic in this case, which will let the optimizer reduce the function to its original shape: uint64_t brightened(uint64_t x, uint64_t y) { return x + y; } LLVM IR before optimizations: define i64 @brightened_0x1000(i64 %0, i64 %1) { entry: %state = alloca %State, align 8 %rdi = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 5 store i64 %0, ptr %rdi, align 4 %rsi = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 4 store i64 %1, ptr %rsi, align 4 %stack = alloca i8, i64 4096, align 1 %2 = getelementptr i8, ptr %stack, i64 4088 %3 = ptrtoint ptr %2 to i64 %rsp = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 6 store i64 %3, ptr %rsp, align 4 store i64 3735928559, ptr %2, align 1 call void @lifted_0x1000(ptr %state, ptr @RAM) %rax = getelementptr inbounds nuw %State, ptr %state, i32 0, i32 0 %4 = load i64, ptr %rax, align 4 ret i64 %4 } After optimizing the module with default<O1>: define i64 @brightened_0x1000(i64 %0, i64 %1) { entry: %2 = add i64 %1, %0 ret i64 %2 } Memory / Stack To handle memory accesses, we create a global RAM variable and pass that to our memory argument. In the previous example it folded away, but we need to handle it separately. The simplest form is access to a pointer parameter: uint64_t lift4_read(uint64_t *n) { return *n ^ 1337; } With our current brightening strategy the lifted code (after optimizations) would look like this: define i64 @brightened_0x1000(i64 %0) { entry: %1 = getelementptr i8, ptr @RAM, i64 %0 %2 = load i64, ptr %1, align 1, !alias.scope !19, !noalias !22 %3 = xor i64 %2, 1337 ret i64 %3 } We need to detect the getelementptr i8, ptr @RAM, i64 %0 shape and replace it with an inttoptr instruction: define i64 @brightened_0x1000(i64 %0) { entry: %1 = inttoptr i64 %0 to ptr %2 = load i64, ptr %1, align 1, !alias.scope !19, !noalias !22 %3 = xor i64 %2, 1337 ret i64 %3 } The stack can be modeled by allocating a local stack variable and pointing rsp to the end of that buffer (since on x86 the stack grows towards lower addresses): uint64_t brightened(uint64_t x, uint64_t y) { uint8_t stack[4096]; State state; state.rdi = x; state.rsi = y; state.rsp = (uint64_t)&stack[sizeof(stack) - 8]; lifted(&state, RAM); return state.rax; } Putting everything together in brighten.py: from llvm import Linkage, Module, Opcode, Value, global_context from bfs import lift_bfs from container import Container, RawContainer OPT_PIPELINE = "default<O1>" def rewrite_ram_geps(module: Module, ram: Value): """Replace GEPs rooted at @RAM with inttoptr(address).""" types = module.context.types for gep in ram.users: if not gep.is_instruction or gep.opcode != Opcode.GetElementPtr: raise ValueError(f"unexpected @RAM user: {gep}") if gep.get_operand(0) != ram: raise ValueError(f"unexpected @RAM GEP base: {gep}") if gep.num_operands == 2: if gep.gep_source_element_type != types.i8: raise ValueError(f"expected i8 ptradd-style @RAM GEP: {gep}") address = gep.get_operand(1) elif gep.num_operands == 3: zero = gep.get_operand(1) if not zero.is_constant_int or zero.const_zext_value != 0: raise ValueError(f"expected zero first @RAM GEP index: {gep}") address = gep.get_operand(2) else: raise ValueError(f"unexpected @RAM GEP shape: {gep}") with gep.create_builder() as ir: ptr = ir.inttoptr(address, types.ptr) gep.replace_all_uses_with(ptr) gep.erase_from_parent() if not ram.users: ram.delete_global() module.verify_or_raise() def define_ret_stub(module: Module): """Make the modeled return hook removable for this demo wrapper.""" ret_handler = module.get_function("__striga_ret") if ret_handler is not None and ret_handler.is_declaration: ret_handler.linkage = Linkage.Internal entry = ret_handler.append_basic_block("entry") with entry.create_builder() as ir: ir.ret_void() def lift_brightened(container: Container, entry: int, args: list[str]): with global_context().create_module("blog") as module: sem = lift_bfs(module, container, entry, verbose=True) # Convenience aliases types = module.context.types i8 = types.i8 i64 = types.i64 # Global RAM array ram = module.add_global(types.array(i8, 0), "RAM") # TODO: support different register sizes brightened_ty = types.function(i64, [i64 for _ in args]) brightened = module.add_function(f"brightened_{hex(entry)}", brightened_ty) with brightened.create_builder() as ir: state = ir.alloca(sem.state_ty, "state") def reg_ptr(name: str) -> Value: return ir.struct_gep(sem.state_ty, state, sem.reg_indices[name], name) # Assign arguments to register state for i, name in enumerate(args): ir.store(brightened.get_param(i), reg_ptr(name)) # Set up function stack stack = ir.alloca(i8, i64.constant(4096), "stack") stack_ptr = ir.gep(i8, stack, [i64.constant(4096 - 8)]) ir.store(ir.ptrtoint(stack_ptr, i64), reg_ptr("rsp")) # Set up return address retaddr_store = ir.store(i64.constant(0xDEADBEEF), stack_ptr) retaddr_store.inst_alignment = 1 # Call lifted function ir.call(sem.function, [state, ram]) # Load return value from rax and return it ir.ret(ir.load(i64, reg_ptr("rax"))) module.verify_or_raise() # 1. Inline/optimize with @RAM assigned to the lifted memory parameter. module.optimize(OPT_PIPELINE) # 2. Brighten lifted memory: @RAM + integer address -> inttoptr(address). rewrite_ram_geps(module, ram) # 3. Now that RAM accesses have been brightened, discard the modeled ret # hook for this demo and let LLVM clean up the remaining wrapper noise. # Undefined flag helpers are already declared memory(none) by Semantics, # so their dead uses fold away without local stub definitions. define_ret_stub(module) module.verify_or_raise() module.optimize(OPT_PIPELINE) print(brightened) This cleanly lifts the following (unoptimized) function: ; 0x1000 push rbp ; 0x1001 mov rbp, rsp ; 0x1004 mov qword ptr [rbp - 8], rdi ; 0x1008 mov rax, qword ptr [rbp - 8] ; 0x100c pop rbp ; 0x100d ret define i64 @brightened_0x1000(i64 returned %0) { entry: ret i64 %0 } Conclusion Hopefully this was an insightful introduction to lifting to LLVM IR. Feel free to check out the repository at LLVMParty/striga and reach out if you do something interesting with it! Note: Striga is not meant to be a production-ready lifter. There are no tests and only a very limited subset of x86 has been implemented. Thanks to the reviewers: Jack Royer Justas Masiulis Peter Goodman eversinc33 x86matthew LLVM IR references: A Gentle Introduction to LLVM IR A Journey to understand LLVM-IR! Mapping High Level Constructs to LLVM IR IR is better than assembly Introduction to LLVM Learning LLVM Part 1, Part 2 LLVM Passes for Security Part 1, Part 2, Part 3, Part 4  ( 32 min )

Learn how Jamf uses native macOS and iOS threat prevention to close security gaps and prove compliance where PC-based tooling leaves Apple devices exposed.  ( 6 min )

2026 Link to heading macOS Exploit Mixtape – Hack Like it’s the 80s /CA: Gergely Kalman/ (Zer0Con) Link to heading Finding Vulnerabilities in Apple Packages at Scale (SecurityFest, MacDevOpsYVR) Presentation Video - SecurityFest 2025 Video - MacDevOpsYVR 2025 The Evolution of macOS Security from the Desert to the Lake (MacSysAdmin, IT Defense 2026, University of Utah MacAdmins meeting)  ( 2 min )

Admin Insights for Windows 365 is a feature, currently in public preview, that surfaces prioritized health and performance signals for your Cloud PCs directly in the Microsoft Intune admin center. Instead of hunting through separate reports, you see dynamically generated insight cards on a single overview page. The feature covers connectivity, provisioning, performance, and utilization issues. It requires Windows 365 Enterprise or Windows 365 Flex licensing and appropriate read permissions. Source

In hybrid Exchange environments, organizations have long been forced to keep an on-premises Exchange Server running just to manage Exchange-related settings for mailboxes already hosted in Exchange Online. Microsoft has been addressing this with the Cloud-Managed Remote Mailboxes feature, and its latest addition — writeback — entered public preview on May 15, 2026. Writeback automatically pushes Exchange attribute changes made in Exchange Online back to your on-premises Active Directory, so internal line-of-business applications that read from AD stay in sync. This article explains what writeback does, what you need to configure it, and how it supports decommissioning your last on-premises Exchange Server. Source

This series of posts focuses on key Active Directory (AD) components that need to be secured in order to ensure AD security is leveled up. In this post, we focus on Group Policy Objects (GPOs) and their permissions. Group Policy provides the ability to change application settings, security settings, install and run code, and more! … Continue reading  ( 6 min )

MSP engineers must resist quick-win scoping in Jamf Pro and build scalable, automated, code-driven deployments instead.  ( 6 min )

MacBook Neo brings powerful, affordable Mac computing to K-12 classrooms — and with Jamf, deploying and managing it alongside iPad has never been easier.  ( 6 min )

Windows Autopatch in the Intune admin center now includes an updated Secure Boot status report that provides device-level visibility into certificate readiness ahead of the 2026 expiry deadline. The report shows which devices have Secure Boot enabled, whether their certificates are up to date, and whether automatic or manual deployment applies. New columns for trust configuration, confidence level, and alerts help you make targeted decisions instead of broad deployments. Source

The May 2026 cumulative update KB5089549 added a new C:\Windows\SecureBoot\ExampleRolloutScripts folder containing seven PowerShell scripts. These scripts are part of Microsoft's sample toolkit for managing Secure Boot certificate migration across enterprise environments. This article explains what each script does, how to run it, and its limitations. Source

Key Takeaways What AI Pentesting Means for Continuous Security Validation Every CISO conversation I’ve had this quarter circles back to the same problem: AI produces more vulnerability findings than security teams can read in a week, and it clouds their understanding of which findings are connected to real business risk. This week’s Wall Street Journal […] The post AI Can Find More Vulnerabilities. Humans Still Decide What Matters. appeared first on Synack.  ( 12 min )

Discover the latest on malicious versions of the pypi package durabletask, matching TeamPCP tactics.  ( 51 min )

Wiz Runtime Sensor support for Google Cloud Run Containers is now generally available, giving teams real-time threat detection and response for their serverless container workloads.  ( 54 min )

Multi-ecosystem supply chain compromise by TeamPCP targets GitHub, NPM, and VSCode to steal credentials and establish persistence.  ( 56 min )

This research was recently presented at BSides Luxembourg 2026. This blogpost documents our findings presented during the talk. The BSides slides are posted here. Today, we’re also releasing the Docker-based playground utilized for the demos so anyone interested can reproduce the findings locally: doyensec/cfitsio-efs-playground. In our previous post on CFITSIO, we wrote about the AI-assisted fuzzing pipeline and the memory corruption issues found in its Extended Filename Syntax (EFS). This was only half of the story. We kept thinking that even without memory issues, EFS seems like a pretty powerful and rather risky feature. The EFS page is full of very interesting use cases. To quote some of them (emphasis mine): ‘rawfile.dat[i512,512]’: reads raw binary data array (a 512x512 short intege…  ( 11 min )

Microsoft has made Platform Single Sign-On (PSSO) during Automated Device Enrollment (ADE) generally available for macOS. The new EnableRegistrationDuringSetup setting in Microsoft Intune completes device registration and SSO configuration automatically during Setup Assistant — the initial macOS setup wizard — before the user ever reaches the desktop. This article explains what PSSO is, why the new setting matters, what you need to configure it, and what limitations to expect. Source

Microsoft Identity Manager (MIM) 2016 Service Pack 3 (SP3) became generally available on May 14, 2026, after an initial release in late March 2026 that Microsoft quietly withdrew without public explanation. SP3 is primarily a platform compatibility update: it adds support for SQL Server 2022, SharePoint Subscription Edition (SE), and Exchange Server SE. The most technically significant additions are Azure SQL Database support for the Synchronization Service using managed identities and claims-based authentication via Active Directory Federation Services (AD FS) for the MIM Portal. MIM 2016 remains supported until January 9, 2029. Source

Learn what seamless learning access means and discover how it supports classroom delivery while simplifying iPad usage.  ( 6 min )

Eliminate cryptographic blind spots and neutralize legacy debt with an integrated cryptographic asset inventory. Identify risks across code, cloud, and runtime, using the Wiz Security Graph to prioritize migration and protect against "Harvest Now, Decrypt Later" attacks.  ( 58 min )

Following two days of intense competition, Day Three of Pwn2Own Berlin 2026 brought the curtain down on an incredible event. Security researchers delivered their final exploits, pushing enterprise systems to the limit one last time as the race for Master of Pwn came to a close. Day Three added to an already historic event, bringing the final totals to $1,298,250 awarded for 47 unique 0-day vulnerabilities across three days of competition. DEVCORE claimed the title of Master of Pwn with a commanding 50.5 points and $505,000 — a dominant performance across all three days. STARLabs SG finished in second with 25 points and $242,500, followed by Out Of Bounds in third with 12.75 points and $95,750. Congratulations to all the researchers who participated, and a special thank you to OffensiveCon …

Cross-platform security tools leave critical blind spots on Apple devices in K-12 environments. Read on to learn why Apple's hardware-rooted architecture demands purpose-built solutions, and what Apple-native security like Jamf Protect actually looks like.  ( 11 min )

Key Takeaways Why Continuous Security Validation Matters California’s evolving privacy regulations are doing more than adding another compliance requirement. They’re changing how organizations think about cybersecurity governance, accountability, and operational resilience. The latest guidance around cybersecurity audits under the California Consumer Privacy Act (CCPA) signals a broader shift happening across the industry: security leaders are […] The post How CCPA Cybersecurity Audits Are Reshaping Cyber Governance appeared first on Synack.  ( 15 min )

Azure Fabric Backdoor With A Twist  ( 15 min )

State of the Art of Private Key Security in Blockchain Ops - 4. Approvals and Policies  ( 13 min )

No content preview  ( 7 min )

David Brauchler III delivers a fascinating Black Hat talk on the root cause of AI-based vulnerabilities and why security architecture is the real solution.  ( 7 min )

MCP Bridge Upgrade  ( 7 min )

Black Hole of Trust: SEO Poisoning in Silver Fox’s Space Odyssey  ( 7 min )

State of the Art of Private Key Security in Blockchain Ops - 3. Private Key Storage and Signing Module  ( 12 min )

The Symbols of Operation code data confusion ada lovelace  ( 6 min )

Public Report: AWS EKS Security Claims  ( 7 min )

Public Report: Google Private AI Compute Review  ( 7 min )

State of the Art of Private Key Security in Blockchain Ops - 2. Common Custody Solutions Architectures  ( 12 min )

Legacy Technology in Transport: More Than “Old Tech”  ( 7 min )

No content preview  ( 14 min )

Concepts, Types of Wallets and Signing Strategies  ( 12 min )

Bridging the Valley of Death: How Assurance Takes Us from Proof of Concept to Minimum Viable Product  ( 7 min )

Goal-Based Regulation  ( 7 min )

Unmasking Techno Sophists  ( 6 min )

Public Report: VetKeys Cryptography Review  ( 7 min )

Your point of departure for forensic readiness - Digital Forensics Incident Response  ( 10 min )

Euro 7, Anti-tampering, and the Expanding Cybersecurity Landscape  ( 6 min )

Discover how ransomware affects patient care, with insights from NCC Group on clinical vulnerabilities and sector trends.  ( 16 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 8 min )

No content preview  ( 6 min )

No content preview  ( 13 min )

This article details a vulnerability in Java Web Start that allows file inclusion through manipulated system properties.  ( 10 min )

No content preview  ( 7 min )

No content preview  ( 9 min )

No content preview  ( 8 min )

No content preview  ( 9 min )

No content preview  ( 7 min )

No content preview  ( 10 min )

Hackproofing Lotus Domino Web Server  ( 6 min )

No content preview  ( 24 min )

No content preview  ( 6 min )

Explore how a command injection flaw in Citrix Access Gateway could allow attackers to execute arbitrary system commands.  ( 9 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 9 min )

No content preview  ( 7 min )

No content preview  ( 10 min )

No content preview  ( 6 min )

No content preview  ( 9 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 11 min )

No content preview  ( 7 min )

No content preview

No content preview  ( 11 min )

No content preview  ( 7 min )

No content preview  ( 11 min )

No content preview

No content preview  ( 30 min )

No content preview  ( 8 min )

No content preview

No content preview

No content preview

No content preview

No content preview

Discover how LDAPFragger uses LDAP attributes to evade network restrictions and exfiltrate data covertly.  ( 15 min )

No content preview

No content preview  ( 11 min )

No content preview  ( 6 min )

No content preview  ( 8 min )

Advanced SQL Injection in SQL Server Applications  ( 6 min )

No content preview  ( 10 min )

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 12 min )

No content preview  ( 7 min )

No content preview  ( 19 min )

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview  ( 7 min )

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview  ( 7 min )

A practical approach to neutralizing Log4j’s JNDI vulnerability without upgrading the entire library.  ( 12 min )

No content preview  ( 20 min )

No content preview  ( 7 min )

No content preview  ( 9 min )

No content preview

No content preview

No content preview

No content preview  ( 8 min )

No content preview  ( 6 min )

No content preview  ( 8 min )

No content preview  ( 6 min )

No content preview  ( 9 min )

No content preview  ( 19 min )

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview  ( 10 min )

No content preview  ( 16 min )

No content preview  ( 17 min )

No content preview  ( 6 min )

No content preview  ( 12 min )

No content preview  ( 7 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 10 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 8 min )

No content preview  ( 7 min )

No content preview  ( 9 min )

Explore the introduction to exploiting CVE-2018-8611 in Windows Kernel Transaction Manager (KTM) with NCC Group’s expert analysis.  ( 23 min )

No content preview  ( 12 min )

No content preview  ( 6 min )

No content preview  ( 12 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 9 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 19 min )

No content preview

This post reveals how a simulated attack demonstrated the ease of breaching hospital systems using basic resources.  ( 13 min )

No content preview  ( 17 min )

Learn how anti-fuzzing techniques enhance defence-in-depth strategies and protect applications from fuzzing-based vulnerabilities.  ( 13 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 7 min )

Explore the remote code execution flaw in ImpressPages CMS and learn best practices for vulnerability remediation.  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 10 min )

Learn how forensic techniques can recover deleted entries from the Windows Registry for investigation and analysis.  ( 7 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 10 min )

No content preview  ( 7 min )

Discover how multivariate cryptography fits into the future of secure communications and what makes it unique among quantum-safe algorithms.  ( 21 min )

No content preview  ( 11 min )

Understand the mechanics behind SMB hash hijacking and user tracking in MS Outlook. Our advisory covers attack vectors, testing methods, and fixes.  ( 12 min )

No content preview  ( 10 min )

No content preview  ( 14 min )

No content preview  ( 12 min )

No content preview  ( 9 min )

No content preview  ( 18 min )

No content preview  ( 8 min )

No content preview  ( 13 min )

Discover how attackers could exploit memory handling flaws in OpenOffice.org to compromise system integrity and user safety.  ( 9 min )

No content preview  ( 9 min )

No content preview  ( 29 min )

No content preview  ( 7 min )

Explore a detailed breakdown of the Bybit hack, uncovering attack methods, vulnerabilities, and security lessons learned.  ( 15 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

Explore known XML-based attack methods including DTD abuse, schema exploits, and entity expansion vulnerabilities.  ( 7 min )

No content preview  ( 7 min )

No content preview  ( 7 min )

No content preview  ( 8 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

Nagios XI Network Monitor is vulnerable to blind SQL injection. Learn the impact, exploitation risks, and mitigation steps.  ( 7 min )

No content preview  ( 6 min )

A technical overview of NFC chip vulnerabilities and protection strategies for secure communication.  ( 15 min )

No content preview  ( 11 min )

Learn why CVE-2021-3156 poses a threat to VMware vCenter and how to protect your infrastructure from attacks.  ( 30 min )

No content preview  ( 9 min )

Explore how Yara can detect Authenticode timestamp anomalies in PE files and enhance malware analysis.  ( 10 min )

No content preview  ( 17 min )

No content preview  ( 7 min )

No content preview  ( 9 min )

No content preview  ( 9 min )

No content preview  ( 18 min )

No content preview  ( 11 min )

No content preview  ( 12 min )

No content preview  ( 12 min )

No content preview  ( 10 min )

No content preview  ( 11 min )

No content preview  ( 25 min )

No content preview  ( 10 min )

No content preview  ( 10 min )

No content preview  ( 16 min )

No content preview  ( 13 min )

No content preview  ( 8 min )

No content preview  ( 9 min )

No content preview  ( 17 min )

No content preview  ( 13 min )

Discover how fuzzing RTSP streams uncovered vulnerabilities in VLC and advanced secure software development.  ( 11 min )

No content preview  ( 15 min )

No content preview  ( 8 min )

No content preview  ( 27 min )

Insights from honeypot research on F5 TMUI RCE vulnerability. Understand attack patterns and steps to strengthen your security posture.  ( 14 min )

Discover how early-stage security planning in hardware and embedded systems can dramatically reduce attack surfaces.  ( 12 min )

No content preview  ( 16 min )

No content preview  ( 15 min )

No content preview

No content preview  ( 10 min )

A brief look at Windows telemetry: CIT aka Customer Interaction Tracker  ( 24 min )

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview

No content preview  ( 7 min )

No content preview  ( 6 min )

No content preview  ( 7 min )

No content preview  ( 7 min )

No content preview  ( 9 min )