OpenAI Unleashes GPT-6 Cyber on Bugs

🔥 What happened OpenAI launched GPT-6 Cyber, a model built to hunt security flaws. It flagged 948 potential issues in the Linux kernel and 81 unknown bugs in Firefox. Access is restricted to vetted orgs via the “Daybreak Red” program. 💡 Why it matters One developer used GPT-5.6 to find a deeply nested bug in two hours that humans missed for years — then patched it via Debian. OpenAI spent $1.2M just on the Linux scan and dangles $1B in bounties for critical infrastructure. The window between discovery and exploit is collapsing. ⚡ Our take OpenAI is arsonist and firefighter in one: the same frontier models that find bugs also lower the bar for attackers. Anyone who thinks patching solves this hasn’t grasped how fast AI moves.