OpenAI launches GPT-6 Cyber with ‘Critical’ cybersecurity rating

OpenAI has introduced GPT-6 Cyber, a cybersecurity-focused model designed to identify and exploit security flaws. This model was released to a select group of vetted customers on Thursday and has been rated as 'Critical' in cybersecurity under OpenAI’s safety framework. The primary purpose of GPT-6 Cyber is to assist security researchers in discovering vulnerabilities, assessing how they could be exploited, and rectifying these flaws before malicious actors do, as OpenAI explained in response to the growing misuse of AI by cybercriminals.

Utilizing GPT-6 Cyber, OpenAI's researchers identified 81 previously unknown Firefox vulnerabilities, which Mozilla has since classified as high severity and released fixes for. Moreover, the model contributed to creating functional exploits for 948 Linux kernel vulnerabilities, including 588 that had not been previously known. OpenAI has also disclosed 304 critical Linux findings to maintainers, along with proposed patches.

In response to the escalated threats, OpenAI has elevated the cybersecurity classification of GPT-6 Cyber to 'Critical' from 'High' for its predecessor GPT-5.6 Cyber. The company is observing an increase in attempts to misuse AI, including attempts by malicious cyber actors and groups attempting to distill its models. GPT-6 Cyber is currently available to a limited number of customers through OpenAI's Daybreak Red access tier, which caters to advanced authorized security work such as penetration testing and exploit development.

OpenAI plans to expand access to GPT-6 Cyber to all approved Daybreak Red customers over the upcoming weeks. The company is also making its earlier GPT-5.6 Cyber model available to all Daybreak Red customers starting Thursday.