OpenAI Releases GPT-6 Cyber for Vulnerability Research
OpenAI launched GPT-6 Cyber, a specialized AI model for vetted security professionals that identified hundreds of previously unknown vulnerabilities in Firefox and Linux.
OpenAI released GPT-6 Cyber on Thursday, a specialized AI model designed to identify and exploit security vulnerabilities. The model outperforms the general-purpose GPT-6 Astra in advanced vulnerability research and exploit development. Due to the risk of misuse, OpenAI is limiting access to a small group of vetted customers through Daybreak Red, an access tier for authorized security work such as penetration testing and red-teaming.
In initial testing, the model identified 81 previously unknown high-severity vulnerabilities in the Firefox browser, which Mozilla has begun fixing. It also developed exploits for 948 Linux kernel vulnerabilities, including 588 previously unknown flaws. OpenAI disclosed 304 critical Linux findings and proposed patches to maintainers. Under its safety framework, OpenAI rated the model as Critical for cybersecurity, an increase from the High rating given to GPT-5.6 Cyber.
The release follows an observed increase in attempts by malicious actors to misuse and distill AI models. The move is also part of a strategy to capture a growing enterprise market, with AI cybersecurity spending forecast to reach $51.3 billion this year. OpenAI faces competition in this sector from Google's Fairwind Program and Anthropic's Glasswing program, both of which provide vetted users with specialized access to frontier models.
