Expanding Daybreak as the Cyber Defense Window Narrows

OpenAI introduces Daybreak Blue and Red access tiers, with GPT-5.6-Cyber completing 95% of advanced cyber requests versus 1.5% for GPT-5.6 Sol.

• Daybreak Blue provides access to GPT-5.6 Sol with reduced guardrails for defensive tasks like vulnerability discovery and incident response.
• Daybreak Red offers GPT-5.6-Cyber, trained to reduce refusals on dual-use tasks like exploit-chain development and zero-day research.
• GPT-5.6-Cyber found two V8 vulnerabilities, including CVE-2026-15903, a high-severity issue fixed by Google.
• Daybreak access requires identity verification and approved-use restrictions; individual accounts must use hardware security keys by September 1, 2026.
• GPT-5.6-Cyber is rated High but not Critical under OpenAI's Preparedness Framework.

Approved cybersecurity defenders can now access models with fewer refusals and better performance on advanced tasks, potentially accelerating defensive research and vulnerability patching.

Sources

• OpenAI: Expanding Daybreak as the Cyber Defense Window Narrows

Read this as text

Back to the AI news