OpenAI opens GPT-6 Cyber to approved security researchers
GPT-6 Cyber improves vulnerability discovery, validation and patch development while adding controls designed to keep testing within authorized scope; access starts with approved Daybreak Red customers.
AT A GLANCE
Key points
- GPT-6 Cyber is initially limited to approved Daybreak Red customers.
- OpenAI reports 48.5% on ExploitGym, versus 32.1% for GPT-5.6 Cyber.
- The model is classified as Critical for cybersecurity under the Preparedness Framework.
What is changing?
OpenAI has introduced GPT-6 Cyber, a model built for cybersecurity research. It focuses on finding software vulnerabilities, validating exploitability and developing candidate patches. Initial access is limited to a select group of approved Daybreak Red customers through the API and Codex.
According to OpenAI’s evaluations, GPT-6 Cyber scored 48.5% on ExploitGym, compared with 32.1% for GPT-5.6 Cyber. The company also says the model avoided attacking out-of-scope services in 99.93% of tests without its deployment safety prompt and 100% with the prompt. These are provider measurements.
The model is classified as Critical for cybersecurity under OpenAI’s Preparedness Framework. Access is restricted with controls including IP allowlisting, misuse monitoring, private safety processing, sandboxing and human oversight. Security teams should validate results only in authorized environments and keep model actions under independent review.
Explore the original sources
Source date: 5 November 2026.
Published by AI Radar: 5 Nov 2026, 21:10 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…