OpenAI details GPT-6 Cyber, limited to approved security users
OpenAI published a system card on November 5 for GPT-6 Cyber, a version of GPT-6 Astra built for authorized security work. It is available only to approved Daybreak Red users and organizations.
What changed
The card covers tasks such as vulnerability discovery, exploit development and web pentesting. It lists the model as Critical in Cybersecurity, High in the Biological and Chemical domain, and below High in AI Self-Improvement.
Why it matters
Stronger security models can help defenders find flaws faster, and the same skills carry misuse risk, which is why access is restricted.
What remains uncertain
The results are OpenAI’s own. It warns that contamination may inflate one benchmark, and says neither model tested solved the 15 highest-difficulty challenges in one evaluation.
What to watch
Watch for independent testing and for any change to who can get Daybreak Red access.
Source
- deploymentsafety.openai.comGPT-6 Cyber System Card
Why this ranks here
A restricted frontier security model has wide implications for defenders, but the evidence so far comes from OpenAI’s own card.
- Impact8
- Reach5
- Novelty7
- Commitment7
- Coverage quality6
- Future potential7