OpenAI has introduced GPT-6 Cyber, a specialised AI model designed to identify software security weaknesses, develop working exploits and help cybersecurity experts fix vulnerabilities before hackers can take advantage of them.
The model was released on November 5, 2026, to a restricted group of approved customers. OpenAI has assigned it a “Critical” cybersecurity rating, highlighting the growing capabilities and potential risks of AI-powered cyber operations.
Unlike general-purpose chatbots, GPT-6 Cyber is specifically designed for advanced security research, including vulnerability discovery, penetration testing and exploit development.
What Can OpenAI’s GPT-6 Cyber Do?
The model’s early security research results indicate how AI could change vulnerability detection.
According to OpenAI, GPT-6 Cyber helped researchers:
- Identify 81 previously unknown Firefox vulnerabilities, which Mozilla classified as high severity. Security fixes are now being released.
- Develop working exploits for 948 Linux kernel vulnerabilities, including 588 that had not previously been known.
- Report 304 critical Linux security findings to maintainers, together with proposed software patches.
These results suggest that AI models could help security researchers examine large and complex software systems much faster, potentially reducing the time between discovering a vulnerability and fixing it.
Why Has OpenAI Given GPT-6 Cyber a Critical Rating?
The “Critical” classification is higher than the “High” cybersecurity rating given to its predecessor, GPT-5.6 Cyber.
The concern is the model’s ability to move beyond identifying software weaknesses and help develop functional exploits. Such capabilities can strengthen cyber defences, but they could also be misused to attack computer systems.
OpenAI says it has observed increasing attempts to misuse AI, including activity by malicious cyber actors and groups attempting to distil its models.
This creates a difficult challenge for AI companies: making powerful security tools available to legitimate researchers without handing similar capabilities to attackers.
Who Can Use GPT-6 Cyber?
OpenAI is initially offering GPT-6 Cyber through Daybreak Red, its restricted access programme for authorised cybersecurity work.
Approved users can access advanced capabilities for activities such as penetration testing and exploit development. OpenAI plans to expand availability to all approved Daybreak Red customers over the coming weeks.
The earlier GPT-5.6 Cyber model is also being made available to all Daybreak Red customers.
Can AI Become a Major Cybersecurity Defence Tool?
GPT-6 Cyber represents a significant development in AI-assisted vulnerability research. It could help organisations discover weaknesses, test software security and prepare patches before vulnerabilities are exploited.
But the same technology introduces a new security risk. As AI becomes better at finding and exploiting weaknesses, controlling access to these capabilities will become increasingly important.
The next cybersecurity battle may involve AI systems defending digital infrastructure against threats powered by AI itself.
