OpenAI launches GPT-6 Cyber, adding an AI model for cybersecurity

OpenAI has launched GPT-6 Cyber, a cybersecurity model reported to carry a “Critical” cybersecurity rating. The supplied coverage does not explain the rating criteria, access conditions or pricing.^[1]

Separate coverage describes a practical example of AI-assisted vulnerability discovery: Max Justicz, cofounder and chief technology officer of robotics startup Charge Robotics, used an OpenAI model to find a severe weakness in Ubuntu, a widely used open-source Linux operating system. If exploited, the weakness could let attackers take control of servers running Ubuntu. The summary does not identify that model as GPT-6 Cyber.^[2]

If models like this can reliably find software weaknesses, security teams could inspect more code with less manual work. Cybersecurity vendors could incorporate them into products or services to expand testing coverage or lower the cost per check; the payoff depends on accuracy, pricing and the cost of fitting them into existing workflows.^[1][2]

Finding more weaknesses also creates follow-up work. Software developers and server operators must verify problems, prioritize fixes and deploy updates, potentially increasing near-term engineering spending. The Ubuntu example shows the value of that work: finding and fixing an exploitable weakness can reduce the risk of server takeover, although the coverage does not quantify savings or added spending.^[2]

Impact