OpenAI Launches GPT-6 Cyber Security Model

It prevents 99.93% of out-of-scope attacks without prompts and reaches 100% with prompts. The model identified 81 previously unknown high-risk vulnerabilities in Firefox and helped Mozilla initiate fixes, while reproducing 1,724 Linux kernel flaws and generating exploit code for 948 of them.

Summarised from the original article by an open-source model, then checked automatically: every number and English name must appear in the source. The original article is authoritative.