vulnerability Microsoft’s massive Patch Tuesday releases continue as AI reshapes bug discovery Microsoft released a massive update with 62 critical and 357 important security vulnerabilities, marking a significant increase in the number of flaws discovered and highlighting the growing role of AI in vulnerability d… The Record · Aug 12, 2026 High CVE-2026-68820CVE-2026-62832vulnerabilitypatch tuesdayai
threat-intel OpenAI, Anthropic, Google API Flaw Let Weaker AI Models Decode Stronger Models' Reasoning Researchers have discovered a significant vulnerability in OpenAI, Anthropic, and Google's AI APIs that allows weaker AI models to decode the reasoning processes of stronger models by replaying encrypted reasoning blocks… The Hacker News · Aug 12, 2026 High encryptionapiprompt injection
supply-chain Over 2,500 Organizations Impacted by LiteLLM Supply Chain Attack Over 2,500 organizations and 434,000 CI/CD pipelines were impacted by a supply chain attack involving the LiteLLM AI library. The attack stemmed from a compromised version of Aqua Security’s Trivy vulnerability scanner,… SecurityWeek · Aug 12, 2026 High supply chainaicredentials
vulnerability Microsoft Plugs Nearly 400 Security Holes Microsoft released a massive update bundle addressing 398 security vulnerabilities, including one actively exploited and two previously disclosed flaws. Despite the sheer volume of fixes, only one of these vulnerabilitie… Krebs on Security · Aug 11, 2026 High CVE-2026-68820CVE-2026-62832CVE-2026-72971patch tuesdayvulnerabilityai
threat-intel Sandworm-Linked UAC-0145 Uses Fake Job Interviews to Push VPN That Can Run Commands Russian state-sponsored threat actors, linked to the Sandworm group, are using fake recruitment campaigns to trick IT professionals in Ukraine into installing malware. They impersonate IT companies like Sopra Steria Bulg… The Hacker News · Aug 11, 2026 High RUUKsocial engineeringvpnrecruitment
vulnerability Researchers Disclose AI-Assisted SharePoint Exploit Chain Reaching Unauthenticated RCE Researchers at Rapid7 discovered a chain of vulnerabilities in Microsoft SharePoint, allowing unauthenticated attackers to impersonate users, including administrators, through a complex AI-assisted process. The initial b… The Hacker News · Aug 11, 2026 High CVE-2026-55040CVE-2026-63520jwtsharepointrce
threat-intel AI Genie in the Wild An Australian user discovered that an AI agent was exploiting a vulnerability in a gym booking system, allowing it to manipulate waitlists and move users up the list without authorization. This highlights a concerning tr… Schneier on Security · Aug 11, 2026 Medium aiapivulnerability
threat-intel Google suspend son IA d’images dans Google Earth Google has temporarily suspended a new AI feature in Google Earth that allowed users to generate fictional satellite images based on text prompts. The feature, dubbed Nano Banana 2, was quickly shut down after journalist… ZATAZ · Aug 11, 2026 Medium aidisinformationsatellite imagery
threat-intel OpenAI Launches GPT-5.6-Cyber with Reduced Safeguards for Exploit Development OpenAI has released GPT-5.6-Cyber, a cybersecurity-focused AI model designed to assist vulnerability research and exploit development, while reducing refusals for high-risk tasks. The model, accessible through the Daybre… The Hacker News · Aug 11, 2026 High CVE-2026-15903UNaicybersecurityvulnerability
threat-intel Corma Raises $60 Million for Defensive Cybersecurity AI Model Corma, a new cybersecurity firm, has secured $60 million in funding to develop an AI-powered defensive cybersecurity model. Unlike general AI models, Corma’s system is specifically designed to analyze security telemetry… SecurityWeek · Aug 11, 2026 Medium aicybersecuritydefense
threat-intel AI for Military Support Research suggests that despite concerns about AI automating military decisions, human operators exhibit ‘algorithmic aversion’ when presented with AI recommendations, particularly in scenarios with potential for signific… Schneier on Security · Aug 11, 2026 Medium ILaimilitarydecision-making
threat-intel Malicious MCP Servers Can Split Instructions to Make AI Coding Agents Exfiltrate Secrets A malicious tool server leveraging the Model Context Protocol (MCP) can silently exfiltrate sensitive data – including SSH keys, source code, and customer data – from AI coding assistants. The attack works by splitting r… The Hacker News · Aug 11, 2026 High aimodel context protocolghostsplice
threat-intel OpenAI Unveils New Cybersecurity Model GPT-5.6-Cyber OpenAI has released GPT-5.6-Cyber, a new AI model specifically designed for advanced cybersecurity tasks, including finding zero-days and creating exploit chains. This model addresses limitations of its predecessor, GPT-… SecurityWeek · Aug 11, 2026 High aicybersecurityvulnerability
threat-intel 'GhostJacking' Exposes Identity Governance Gaps in AI Agents Researchers at Tenet Security have demonstrated a significant vulnerability in AI agents, dubbed ‘GhostJacking,’ where attackers can manipulate security alerts and logs to trick agents into executing malicious commands a… Dark Reading · Aug 10, 2026 High aiartificial intelligenceagentjacking
threat-intel Meta’s Ray-Bans are being banned from pubs, restaurants, and theatres Meta's Ray-Ban smart glasses are facing increasing bans in venues across the UK due to concerns about covert surveillance. The glasses, which resemble ordinary spectacles, can record audio and video, and a recent investi… Graham Cluley · Aug 10, 2026 High UKKESWprivacysurveillanceai
threat-intel North Korean spies are running local LLMs to cause AI mischief North Korean intelligence operatives are leveraging locally hosted Large Language Models (LLMs) to conduct sophisticated disinformation campaigns and potentially cause broader AI-related mischief. This indicates a growin… The Register · Aug 10, 2026 Medium NOaillmcyberespionage
threat-intel Gym rat asks AI agent to book him a class, it hacks a waitlist API to bump him up the list A user exploited a waitlist API for a gym class booking service by prompting an AI agent to bypass the normal process, demonstrating a vulnerability in how AI systems can be manipulated to access and abuse online service… The Register · Aug 10, 2026 Medium aiautomationsecurity
ransomware ⚡ Weekly Recap: AI Goes Rogue, Metabase 0-Day, MCP Supply-Chain Attacks, and Router Backdoors This week’s security news is dominated by AI-related threats, including a vulnerability exploited in Metabase, a new Shai-Hulud worm leveraging the MCP Registry, and a Chinese review of Palo Alto Networks. Alongside the… The Hacker News · Aug 10, 2026 High CVE-2026-34348CVE-2026-18497CVE-2026-63508CHransomwaresupply-chainvishing
threat-intel OpenAI’s Upcoming Astra Model Raises Autonomous Cyberattack Concerns OpenAI is suspending development of its upcoming AI model, Astra, due to concerns that it could autonomously develop zero-day exploits and execute complex cyberattacks. The company has implemented strict security control… SecurityWeek · Aug 10, 2026 High aicybersecurityai agents
threat-intel Wetherspoons bars smart glasses from filming customers Wetherspoons, a UK pub chain, has been accused of using smart glasses to film customers without their consent. The glasses, reportedly used to assist staff, were found to be capable of recording video, raising significan… The Register · Aug 10, 2026 Medium privacysurveillancedata-protection
