OpenAI releases GPT-6 Cyber to vetted security teams, rated 'Critical' for cyber risk

What happened

Initial access is through the API and Codex, for authorised security testing and exploit development.

OpenAI rates GPT-6 Cyber as Critical for cybersecurity under its Preparedness Framework. GPT-5.6 Cyber was rated High.

OpenAI reports a vulnerability recall of 40.3% for GPT-6 Cyber, compared with 32.0% for GPT-6 Astra. These are vendor figures.

Access is set to widen to all Daybreak Red customers in the following weeks.

Why it matters for creators

Gated releases of high-risk models show labs splitting capabilities by use case. The same pattern appears in restricted video and avatar features.

GPT-6 Cyber itself is not a creative tool, and most creators will never use it. The broader point is that labs increasingly release their most sensitive capabilities only to vetted groups, while offering general versions more widely. Creators already see this with some likeness, voice and avatar features that require verification or approval before use.