OpenAI releases cybersecurity model GPT-6 Cyber to vetted security teams

OpenAI has released GPT-6 Cyber, a specialized cybersecurity model, to vetted security teams to discover, test software vulnerabilities, and assist in drafting fixes. Made available to select customers through the Daybreak Red initiative, the model has already identified 81 previously unknown critical vulnerabilities in Firefox, developed exploits for 948 vulnerabilities in the Linux kernel, and disclosed 304 critical issues.