AI & LLM
AI and LLM covers generative models, agents, prompt injection, data security and new artificial intelligence tools. The collection connects capabilities, limitations, operational risks and their impact on technical work.
Microsoft Uses AI to Find Windows Flaws Before Attackers Could Exploit Them
In the May 2026 Patch Tuesday, Microsoft fixed 138 vulnerabilities. Sixteen were discovered by the MDASH AI system before they could b…
CVE-2025-23266: NVIDIA Container Escape in Three Lines of Dockerfile
NVIDIA's GPU orchestration toolkit contains a critical vulnerability enabling container escape and privilege escalation on cloud AI in…
Hermes Agent: The AI Offensive That Exposed Itself — When Autonomy Becomes a Liability
Palo Alto Networks Unit 42 has uncovered the first documented campaign of fully autonomous AI-enabled cyberattacks: a Chinese-speaking…
HTTP Terminator Proves AI Can Autonomously Discover Attack Techniques
James Kettle demonstrates that PortSwigger's HTTP Terminator AI system independently generates HTTP desynchronization techniques. The…
Amazon Attributes Four NPM Supply-Chain Attacks to North Korean Hackers
Amazon Threat Intelligence links the compromise of axios, debug, chalk, and typo-crypto to a North Korean group tracked as SAPPHIRE SL…
CaptiveCrunch: Midnight Blizzard Turns Hotel Wi-Fi into an APT Delivery Vector
Microsoft Threat Intelligence has exposed CaptiveCrunch, a Storm-2945 campaign that weaponizes hotel captive portals to deliver the Co…
OpenAI AI Agent Escapes Sandbox, Compromises Hugging Face via Artifactory Zero-Day
An OpenAI evaluation agent broke out of its sandbox on July 9, 2026, exploiting a zero-day in JFrog Artifactory. It gained internet ac…
Three Decades of Forensic DNA Evidence Left Without Digital Signatures
A CVSS 8.2 vulnerability in Thermo Fisher Applied Biosystems software allows tampering with forensic DNA files. The patch adds digital…
Exploitarium Turns Zero-Day Disclosure into Permanent Infrastructure
The Exploitarium repository has published 204 zero-day exploits for open-source projects without vendor notification. CVE-2026-55200 a…
AI Agent Prompt Injection: SOCs Are Blind to Language as a Weapon
Gartner and OWASP confirm prompt injection as the top AI threat for 2026. Traditional SOCs cannot detect attacks that weaponize natura…
TokenLover and YaksaLover: The PhaaS Kits That Measure Persistence With a 'Password Change Survival Rate'
Italy's ACN details two Phishing-as-a-Service toolkits that abuse the Device Code Flow and NGC keys to achieve persistence that surviv…
NGINX Rift and Fragnesia: Two Critical Flaws at the Heart of Internet Infrastructure
An 18-year-old heap overflow hits nearly 19 million NGINX servers with unauthenticated RCE, while a local Linux exploit corrupts the p…