🇹🇼 Chinese, 5 November 2026, AI-powered security
OpenAI Releases GPT-6 Cyber, a Stronger Network Security Model
GPT-6 Cyber has discovered 81 high-risk vulnerabilities in Firefox and provided testing evidence and reproduction plans.
OpenAI has released GPT-6 Cyber, a network security model that enhances software vulnerability discovery, verification, and exploitation development capabilities. In Firefox security research, GPT-6 Cyber found 81 previously undiscovered high-risk vulnerabilities and provided corresponding testing evidence and reproduction plans.
OpenAI reported the issues to Mozilla and submitted repair suggestions. Mozilla has since released patches for several of the vulnerabilities.
GPT-6 Cyber has also successfully reproduced 1724 vulnerabilities in the Linux kernel and developed exploitable programs for 948 of them.