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

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

Argentina is considering a new legal framework to attract artificial intelligence companies by establishing a category for non-human corporations. These entities would be operated by AI agents or robots and would not require human shareholders to function. The proposal aims to provide these systems with limited liability protections to manage risks associated with independent machine judgment.

Source

The European Commission has rejected Apple's request for an 18-month exemption from the Digital Markets Act (DMA) regarding the rollout of its upgraded Siri AI. Regulators stated that the decision to delay the feature in the European Union rests solely with Apple, dismissing claims that EU rules prevent the launch. The DMA is a regulatory framework designed to ensure fair competition by requiring major tech platforms to provide rivals with equal access to their ecosystems.

Source

Microsoft has released Windows 11 Insider Build 26220.8575 to the Beta channel for version 25H2. The most significant addition in this release is the ability to extend update pauses indefinitely. This change allows users to bypass previous limitations on how many times they could delay system updates.

Source

Frontier AI models accelerate the speed of vulnerability discovery and exploit adaptation, allowing attackers to bypass traditional signature-based defenses more effectively. While these models do not change the fundamental stages of an intrusion, they enable the rapid generation of thousands of exploit variations. Security teams must shift focus from the speed of patching to the robustness of the architecture surrounding potential vulnerabilities.

Source

Researchers have developed a proof-of-concept computer worm that utilizes locally hosted, open-weight large language models to autonomously navigate and compromise networks. Unlike traditional malware that relies on hardcoded exploits, this agent reasons through target environments and generates tailored attack strategies at runtime. By operating entirely on local infrastructure, the worm bypasses commercial AI safety filters and remains functional even without internet access or vendor API keys.

Source

OpenAI has submitted confidential paperwork to the Securities and Exchange Commission for an initial public offering. This move allows the company to undergo regulatory review of its financial data before making the information available to the public. While a specific date has not been set, the company is currently valued at approximately $852 billion following recent funding rounds.

Source

A new methodology called loop engineering is shifting the focus of software development from manual prompting to the design of autonomous systems. Unlike traditional workflows where an individual prompts an agent for each task, loops involve setting a specific trigger and a verifiable goal for the AI. Once initiated, the agent independently iterates and refines code until the predefined end state, such as passing all unit tests, is successfully achieved.

Source

Anthropic leadership describes a management approach where technical vision is balanced with day-to-day operational execution. The company maintains a focus on ethical development and business utility over purely financial or competitive metrics. This strategy includes a philosophy of fiscal responsibility regarding high capital costs for compute infrastructure to avoid overextending resources.

Source

Check Point has released emergency hotfixes for a critical authentication bypass vulnerability affecting its VPN and firewall products. Tracked as CVE-2026-50751, the flaw stems from a logic error in how the legacy Internet Key Exchange version 1 (IKEv1) protocol validates certificates. This weakness allows unauthenticated attackers to establish remote access sessions and enter the network without providing a valid password.

Source

Anthropic has demonstrated that its Mythos Preview model can autonomously develop working exploits for recently patched vulnerabilities in just hours or minutes. This research shifts the focus from zero-day threats to "N-day" vulnerabilities, where the public release of a security patch provides a roadmap for attackers to reverse-engineer the original bug. By utilizing techniques like patch diffing, the AI can identify software changes and generate code to weaponize flaws that remain present on unpatched systems.

Source

Linux kernel contributors are now utilizing AI-assisted coding to maintain and refactor legacy hardware drivers. Recent updates to the R600 Gallium3D driver involved dozens of commits generated with the help of GitHub Copilot. These efforts specifically target the shader compiler code to ensure continued functionality for AMD hardware released between 2007 and 2010.

Source

Microsoft 365 has introduced Baseline Security Mode, an opt-in feature that bundles 18 secure-by-default configuration settings across authentication, files, and room devices. While this centralized approach simplifies security management, it can conflict with existing environments that rely on legacy authentication protocols or custom meeting room hardware. Administrators should view this mode as a migration target rather than a simple toggle, as enabling it without preparation can disrupt critical business workflows.

Source

Google Chrome is entering the final phase of deprecating Manifest V2 extensions in favor of the newer Manifest V3 standard. Developers have confirmed the removal of internal flags that previously allowed users to bypass restrictions on older browser add-ons. This change effectively disables popular extensions like the original uBlock Origin that rely on the legacy framework.

Source

Recent software and firmware updates from Dell and HP have caused significant stability issues for Windows 11 users. Dell identified that its SupportAssist and Alienware SupportAssist recovery tools were triggering blue screen errors and infinite reboot loops. While initial workarounds suggested uninstalling the software, the manufacturer has now released a specific patch to address the underlying conflict.

Source

Microsoft has established a dedicated Beta channel for Windows 11 version 26H1 to provide a more stable testing environment. This release is specifically engineered as a platform-focused update to support upcoming silicon, including Qualcomm Snapdragon X2 and NVIDIA RTX Spark processors. The new structure separates the Beta channel into the 28000 build series while moving Experimental devices to the 28100 series.

Source

Microsoft has officially acknowledged that the Windows 11 File Explorer performs slower than previous versions and is developing system-level optimizations to address these issues. The primary focus of current internal testing is a significant improvement to bulk file deletion speeds. Initial benchmarks indicate that these changes could make batch deletion operations at least 30% faster for most users.

Source

Russian-aligned threat actors are actively exploiting a path traversal vulnerability in WinRAR to target organizations in Ukraine. The flaw, tracked as CVE-2025-8088, allows attackers to write malicious files outside the intended extraction directory using NTFS Alternate Data Streams. Despite a patch being available since July 2025, unmanaged software installations remain a primary entry point for cyber espionage.

Source