OpenAI unveils GPT-5.6-Cyber in Daybreak to aid cybersecurity teams
Technology
OpenAI is unveiling GPT-5.6-Cyber, a new AI model designed to give cybersecurity teams a serious boost against advanced hacks like exploit chains and privilege escalation.
It's part of the Daybreak program, which helps trusted defenders access smarter tools, while keeping things safe with strict checks to prevent misuse.
Daybreak Red uses GPT-5.6-Cyber scoring 95%
Daybreak now comes in two flavors: Blue (with GPT-5.6 Sol) and Red (featuring the new GPT-5.6-Cyber).
In tests, GPT-5.6-Cyber nailed 95% of tough security prompts, way ahead of previous models.
Big names like IBM, Cisco, and CrowdStrike can use these tools, but only after passing tight identity checks so only legitimate users get access.