OpenAI releases GPT-6 Cyber on Nov 5, 2026, reports 1,724 reproduced vulnerabilities and 948 working exploits, rates it Critical for cybersecurity
OpenAI published GPT-6 Cyber on Nov 5, 2026, reported 1,724 reproduced vulnerabilities and 948 working exploits, rated it Critical for cybersecurity, and released it to a small group of vetted customers.
- Urgency
- L4
- 72
- 80
- 60
- 50
- 85
At a glance
On Nov 5, 2026 OpenAI announced GPT-6 Cyber and published a system card reporting the model reproduced 1,724 vulnerabilities and developed 948 working exploits in testing. OpenAI classified the model as "Critical" in cybersecurity under its Preparedness Framework and said it released the model to a small group of vetted customers. The release raises near-term operational and policy questions about access controls, patch prioritization, and potential misuse risks.
- OpenAI published GPT-6 Cyber on Nov 5, 2026, reported 1,724 reproduced vulnerabilities and 948 working exploits, rated it Critical for cybersecurity, and released it to a small group of vetted customers.
CHRONOS Wire · November 6 · Alert 8
What happened
OpenAI announced a new model called GPT-6 Cyber on Nov 5, 2026. In its tests the model reproduced 1,724 vulnerabilities and produced 948 working exploits, and OpenAI rated it "Critical" in cybersecurity under its internal framework. OpenAI says it released the model to a small group of vetted customers.
Why it matters
High-capability cyber models can change the speed and scale of vulnerability discovery and exploit development; defenders and infrastructure operators need immediate situational awareness and prioritization guidance.
CHRONOS Bottom Line
OpenAI publicly released GPT-6 Cyber on Nov 5, 2026, published quantified test results showing substantially higher exploit generation versus prior models, rated the model "Critical" for cybersecurity, and limited access to a small set of vetted customers, raising near-term questions for defenders and policymakers.
Full detail
What changed
On Nov 5, 2026 OpenAI published a system card and product announcement for a new model, GPT-6 Cyber, that reports 1,724 reproduced vulnerabilities and 948 working exploits in testing and that the model is rated "Critical" for cybersecurity under OpenAI's Preparedness Framework. OpenAI also stated the model was released to a small group of vetted customers.
What is genuinely new
The GPT-6 Cyber model release itself, OpenAI's published system card numbers (1,724 reproduced vulnerabilities and 948 working exploits), and the model's "Critical" cybersecurity rating were first posted by OpenAI and reported by wires on Nov 5, 2026.
Detailed analysis
OpenAI published a system card showing quantitative test results (1,724 reproduced vulnerabilities and 948 working exploits) and classified GPT-6 Cyber as "Critical" under its Preparedness Framework. Reuters wire reporting the same day noted OpenAI released the model to a small group of vetted customers. The published performance numbers show a substantial increase in exploit-generation capability compared with prior cyber-focused models, which has implications for prioritization of patching and defensive resourcing where the model is used. There is no published evidence as of the announcement that the model has been leaked or used maliciously.
U.S. Impact
U.S. Impact assessment pending.