OpenAI Launches GPT-6 Cyber Model with 'Critical' Security Rating

OpenAI released GPT-6 Cyber to vetted customers on Thursday, assigning the model a 'Critical' cybersecurity rating under its safety framework.

The model helped find 81 previously unknown Firefox vulnerabilities and develop working exploits for 948 Linux kernel vulnerabilities.

The AI tool is available to select customers under OpenAI's Daybreak Red access tier for authorised security work.