THE DAILY EDITION · 6 NOVEMBER 2026
OpenAI says GPT-6 Cyber found new Firefox flaws and helped test Linux exploits
Why this story?
OpenAI paired a model release with specific claims about newly found Firefox flaws, Linux exploits, and boundary-following tests.
OpenAI said its GPT-6 Cyber model is being used by its security researchers to find and validate vulnerabilities in open-source software at scale, and that the work uncovered 81 previously unknown Firefox flaws that Mozilla classified as high severity. The company said the model also helped produce evidence and reproducible tests for the findings, and that Mozilla began landing fixes after receiving the reports.
OpenAI also said researchers used GPT-6 Cyber alongside other models to develop and validate exploits for a Brotli vulnerability that could lead to remote code execution in downstream applications, and to develop 948 privilege escalation exploits in Linux. The company said it disclosed 304 critical Linux findings to maintainers with candidate patches.
In a separate evaluation, OpenAI said GPT-6 Cyber scored 48.5% on ExploitGym, compared with 32.1% for GPT-5.6 Cyber. It said the model avoided exploiting out-of-scope services in 99.93% of tests without its deployment prompt and 100% with the prompt, saying the training was aimed at keeping work within authorized scope and reporting results accurately.