OpenAI Launches GPT-6 Cyber to Help Detect and Fix Dangerous Software Vulnerabilities
The new cybersecurity-focused AI model can identify previously unknown security flaws, develop working exploits and help researchers strengthen digital infrastructure, but access is initially restricted to approved security professionals.
OpenAI has introduced GPT-6 Cyber, a specialised artificial intelligence model designed to help cybersecurity researchers discover, investigate and fix software vulnerabilities before malicious hackers can exploit them. Announced on November 5, 2026, the model represents a significant improvement over its predecessor, GPT-5.6 Cyber, with stronger capabilities in vulnerability discovery, penetration testing and exploit development. OpenAI says the model is already helping security teams protect critical software systems.
According to OpenAI, GPT-6 Cyber achieved a 48.5 per cent score on its ExploitGym benchmark, compared with 32.1 per cent for GPT-5.6 Cyber. The company also reported that the model helped researchers identify 81 previously unknown high-severity vulnerabilities in Mozilla Firefox. Additionally, it developed working exploits for 948 Linux kernel vulnerabilities, including hundreds that had not previously been discovered. Mozilla has already begun releasing security fixes for several of the reported Firefox vulnerabilities.
Despite its powerful capabilities, OpenAI has introduced strict safeguards to prevent the technology from being used for unauthorised cyberattacks. The company classified GPT-6 Cyber as Critical in cybersecurity under its Preparedness Framework, reflecting its advanced ability to discover and exploit security weaknesses. Internal evaluations also showed that the model avoided exploiting systems outside its authorised testing scope in 99.93 per cent of tests without additional deployment prompts and 100 per cent with them.
GPT-6 Cyber is initially available to a limited group of verified customers through OpenAI’s Daybreak Red programme, which supports advanced and authorised cybersecurity research. Access will gradually expand to other approved security professionals in the coming weeks. The launch highlights OpenAI’s growing investment in cybersecurity, as increasingly capable AI systems create both new opportunities for digital protection and additional risks if they fall into the wrong hands.
