OpenAI has expanded its Daybreak program for verified cybersecurity professionals, splitting access into two tiers: the basic Blue tier opens GPT-5.6 Sol with system filters removed (for code review, malware analysis, and vulnerability hunting), while the elite Red tier unlocks GPT-5.6-Cyber, purpose-trained on exploit chain creation, authentication bypass, and privilege escalation. The uncensored model has already proven itself in the field, uncovering a critical zero-day (CVE-2026-15903) in Google Chrome’s V8 JavaScript engine that allows arbitrary code execution with browser sandbox escape, plus hundreds of bugs in popular OS kernels and databases. Access is limited to verified enterprise clients and professionals; starting September 1, 2026, all Daybreak accounts must be protected with hardware security keys, and high-risk Codex agent tool calls will require mandatory manual confirmation via Auto-review mode.

Related: OpenAI Expands Cybersecurity Efforts with GPT-5.5-Cyber and Patch the Planet, OpenAI Releases GPT-5.5-Cyber and Expands TAC Program, OpenAI Previews GPT-5.6 Sol

OpenAI — Expanding Daybreak as the Cyber Defense Window Narrows