OpenAI ships GPT-6 Cyber — a hacker AI for vetted defenders only

OpenAI introduced GPT-6 Cyber on November 5, a cybersecurity-specialized version of GPT-6 Astra, available first through the API and Codex to a select group of approved Daybreak Red customers — vetted defenders doing authorized security work.

The model scores 48.5 percent on ExploitGym, up from 32.1 percent for GPT-5.6 Cyber, and OpenAI rates it Critical in cybersecurity under its Preparedness Framework, the top level; it also stayed within authorized scope in 99.93 percent of honeypot tests without deployment prompts, and 100 percent with them.

It is the model behind the 948 Linux exploits: OpenAI says GPT-6 Cyber reproduced 1,724 kernel vulnerabilities, developed working exploits for 948 of them, and found 81 previously unknown high-severity flaws in Firefox, with Mozilla already landing fixes.