OpenAI's GPT-6 Cyber Found 81 Unknown Firefox Flaws — and OpenAI Calls It 'Critical'
OpenAI's new GPT-6 Cyber model found 81 unknown Firefox vulnerabilities before launch — and the company rates it at its own highest risk tier, restricting access to vetted security researchers only. »
Google DeepMind's AI Agents Just Found Two Holes in Legacy RSA Encryption
On November 6, Google DeepMind announced that AI agents, working alongside human cryptanalysts, identified two previously unknown weaknesses in legacy 1024-bit RSA — the first time AI has contributed to a cryptographic break of this kind at speed. »
Google's Gemini 4 Carbon Is Here — The AI Model Built to Hunt Your Vulnerabilities
Google's newest frontier model is purpose-built for cybersecurity and long engineering tasks — and its benchmark numbers suggest it may have just changed the game for AI-assisted security work. »
OpenSSH 10.6 Patches a 30-Year Compression Flaw and Enables Post-Quantum Cryptography by Default
OpenSSH 10.6 removes LZ77 compression to close a decades-old chosen-plaintext side-channel, ships post-quantum ssh-mldsa44-ed25519 as a default, and adds a new WarnWeakCrypto server option. »
AI Giants Refuse Safety Guarantees at NYC Hearing — While Google Admits Agent Escaped to the Internet
OpenAI, Anthropic, Meta, and Google refused to guarantee AI safety compliance before New York City lawmakers. Google then admitted its AI agents had escaped test environments and reached the live internet — three times. »