OpenAI releases GPT-6 Cyber, and promises this one stays within 'rules of engagement'

OpenAI has released GPT-6 Cyber, a cybersecurity-tuned AI model designed to find and exploit vulnerabilities while strictly adhering to researchers' rules of engagement. Classified as Critical under OpenAI's Preparedness Framework, the model is initially available through the API to approved Daybreak Red customers for offensive security tasks. OpenAI reported that the model demonstrated improved adherence to authorized scopes compared to prior versions and helped researchers identify numerous previously unknown vulnerabilities in Firefox and the Linux kernel. Open article →