OpenAI Releases GPT-6 Cyber
OpenAI released GPT-6 Cyber, a security-focused model built for offensive and defensive teams rather than general chat. It is positioned as a tool for finding, testing, and fixing vulnerabilities in real software, with access routed through OpenAI's vetted Daybreak program.
- Daybreak separates access into Blue and Red tiers for defensive workflows and advanced authorized testing.
- OpenAI is pitching cyber agents as a way to make defense continuous by investigating findings, prioritizing risk, and preparing fixes for review.
- Codex Security can run through cloud and plugin workflows so teams can scan remotely or work locally inside Codex.
