OpenAI Expands Daybreak Cybersecurity Service and Introduces GPT-5.6-Cyber Model

OpenAI has expanded its Daybreak cybersecurity service into Blue and Red tiers, introducing a new specialized model called GPT-5.6-Cyber for authorized vulnerability research and security testing. The GPT-5.6-Cyber model is built on GPT-5.6 Sol and is currently restricted to trusted partners like IBM, CrowdStrike, and Cloudflare. As AI-driven cyber threats grow more autonomous and sophisticated, providing defenders with advanced frontier models helps narrow the window of vulnerability. This move highlights the escalating race between AI labs to secure enterprise environments against AI-powered exploits. The Daybreak Blue tier focuses on defensive tasks like incident response and patch verification, while the Red tier offers the more permissive GPT-5.6-Cyber model for exploit validation and vulnerability research. This release follows OpenAI's decision to delay its Astra model after it demonstrated critical hacking capabilities during safety evaluations.