OpenAI News·· 14 h ago
GPT‑6 Cyber: Advancing the frontier of aligned cybersecurity agents
GPT‑6 Cyber: Advancing the frontier of aligned cybersecurity agents
AI summary
Introducing GPT‑6 Cyber, OpenAI’s cybersecurity model for finding exploitable vulnerabilities and developing fixes, trained to stay within authorized scope.
Selection record
Threshold 60Official, first-handFirst 72Second 78
AdmittedSum of both 150 ≥ twice the threshold 120
- Source tier
- Official, first-hand; this tier's threshold is 60
- Pre-filter
- passed：OpenAI发布网络安全AI模型GPT-6 Cyber
- Why it was chosen
- OpenAI has released a dedicated cybersecurity model that stresses staying within authorised bounds, showing how safety limits are set for security agents.
A model scores each item twice, independently, against one written standard, out of 100. An item is admitted only when the two scores add up to twice the threshold. The threshold is set per source tier.
Source: OpenAI News · openai.com