OpenAI has launched GPT-6 Cyber, upgrading its cybersecurity capabilities rating from "High" in GPT-5.6 Cyber to "Critical." According to reports, the model has identified 81 previously unknown high-risk Firefox vulnerabilities and assisted in identifying and developing usable exploits for 948 Linux kernel vulnerabilities, including 588 previously unknown issues. GPT-6 Cyber is currently being rolled out to select Daybreak Red customers, with plans to expand access in the coming weeks. At the same time, OpenAI warned that attempts by malicious actors to abuse cybersecurity models are increasingly on the rise.