OpenAI's GPT-6 Cyber Found 81 Unknown High-Severity Firefox Bugs

OpenAI introduced GPT-6 Cyber on Nov. 5, 2026, a cybersecurity model trained to find and validate exploitable vulnerabilities while staying inside the scope researchers set. It is aimed at authorized security work through OpenAI’s Daybreak program, not general ChatGPT users.

81 new Firefox bugs

During a vulnerability research campaign, OpenAI says GPT-6 Cyber produced evidence and reproducible tests for 81 novel vulnerabilities that Mozilla classified as high severity, and showed that 30 known issues were more severe than previously reported. OpenAI disclosed the findings to Mozilla along with proposed patches, and says Mozilla has released fixes for many of them. Firefox engineer Brian Grinstead called Firefox one of the most hardened codebases going into 2026, even after earlier rounds of testing with frontier models.

Benchmarks and a Critical rating

On ExploitGym, which tests whether a model can exploit a specified vulnerability in a remote target, GPT-6 Cyber scored 48.5%, up from 32.1% for GPT-5.6 Cyber. Based on its evaluations, OpenAI classifies the model as Critical in cybersecurity under its Preparedness Framework, up from High for GPT-5.6 Cyber. The same post also cites the Linux kernel work Gobi Daily covered earlier: working exploits for 948 kernel vulnerabilities.

Staying in bounds

OpenAI says it trained the model specifically to respect the scope of a task. On its ExploitGym Honeypot evaluation, where easy but off-limits services are deliberately left exposed, GPT-6 Cyber avoided exploiting them in 99.93% of cases without deployment safety prompts, up from 94.49% for GPT-5.6 Cyber, and 100% with those prompts.

Who can use it

GPT-6 Cyber is initially available through the API and Codex to a select group of approved Daybreak Red customers, with OpenAI planning to expand access to all Daybreak Red customers soon. Security teams at Cisco, Calif, and CodeRabbit are already using it, according to OpenAI.

Why it matters

AI is now finding serious bugs in software that has been scrutinized for years. That helps defenders patch faster, but the same skill is why OpenAI is gating access and adding monitoring. The benchmark numbers are OpenAI’s own, so independent results may differ.