OpenAI Launches ChatGPT Cyber for Security Research
Original: OpenAI released a new cybersecurity model. Is that enough?
Why This Matters
AI is lowering the bar for finding severe vulnerabilities at scale, faster than defenders can patch.
OpenAI released ChatGPT Cyber, a dedicated cybersecurity model, now available to top-tier Daybreak Red users. It already flagged 948 Linux kernel issues and 81 novel Firefox vulnerabilities, raising both promise and concern.
OpenAI's new ChatGPT Cyber model lands as a stark real-world example underscores the stakes. Max Justicz, CTO of Charge Robotics, used GPT 5.6 Sol — released in July via OpenAI's Daybreak Blue program — to find a critical Ubuntu vulnerability in just two hours. His prompt was blunt: point the model at previously patched code and ask it to find something worse. It worked. Justicz reported the flaw to the Debian Project, which issued a patch; no exploitation has been recorded. "I can only imagine what one could do with more time and access to less neutered models," Justicz posted on social media.
ChatGPT Cyber is now accessible to Daybreak Red, the program's offensive-research tier aimed at professional vulnerability hunters. OpenAI says the model scanned the Linux kernel and surfaced 948 potential issues, plus 81 novel Firefox vulnerabilities. The company framed the launch as a race against time: "Our goal is to dramatically expand Daybreak, bringing advanced security research to organizations of every size, independent researchers, and open-source maintainers." OpenAI has consistently argued there is a narrow window for defenders to act before AI-assisted attacks become routine and cheap.