OpenAI's GPT-6 Cyber finds 81 critical Firefox flaws
In short
OpenAI’s GPT-6 Cyber found 81 critical Firefox flaws and built 948 Linux exploits. It also showed better results in penetration testing benchmarks.
The report matters because stronger cyber tools can help security testing, but they can also be misused.
What the reporting says
OpenAI’s GPT-6 Cyber found 81 critical Firefox flaws and developed 948 Linux exploits. The model also showed improved performance in penetration testing benchmarks.
Why it matters
The model also improved on penetration testing benchmarks. The report frames the result as a cybersecurity capability gain while noting concerns about misuse of advanced tools. No additional technical detail, timing, or downstream action is provided in the source summary.