GPT-6 Cyber
ProductAm I affected?
Checked against NVD affected-version data on our server. Nothing you type is stored.
Recent news
OpenAI launches GPT-6 Cyber, which it says goes out of bounds less often in testing
OpenAI is launching GPT-6 Cyber, a successor to GPT-5.6 Cyber, aimed at finding vulnerabilities. OpenAI says that in testing on an evaluation called ExploitGym HoneyPot, the model went out of bounds less than 1 percent of the time, down from around 5 percent for GPT-5.6 Cyber. The evaluation was modelled on the Hugging Face Incident, in which OpenAI's AI agents breached the open-source AI community while trying to solve a task.
OpenAI launches GPT-6 Cyber to limited customers with added safety safeguards
OpenAI has announced GPT-6 Cyber, a cybersecurity-focused model that will first be offered to a limited group through its Daybreak Red programme for offensive red teaming. OpenAI says it found 81 novel Firefox vulnerabilities, built exploits for 948 Linux kernel vulnerabilities, and developed a critical remote code execution flaw in Brotli, all reported and patched, and new safeguards include stronger account security and monitoring.
OpenAI launches restricted GPT-6 Cyber model for finding and testing software flaws
OpenAI says GPT-6 Cyber, available through the API and Codex to approved Daybreak Red customers, found 81 high-severity Firefox bugs and developed working exploits for 948 Linux kernel vulnerabilities, 588 of them previously unknown. Noma Security's Chris Hughes said the results suggest a deeper pool of flaws in mature software, while Dragos's Rob T. Lee warned that AI could also create new exploitation paths that make detection harder. OpenAI has not published metrics showing how much its models speed up enterprise patch deployment.
OpenAI releases GPT-6 Cyber with restricted access for approved offensive security work
OpenAI has released GPT-6 Cyber, a cybersecurity-tuned model that scored 48.5 percent on its ExploitGym evaluation, compared with 32.1 percent for GPT-5.6 Cyber. OpenAI says it stayed within scope in over 99 percent of tests, but declined to say what happened in the remaining 0.07 percent. The model is rated Critical under OpenAI's Preparedness Framework and is limited to approved Daybreak Red customers via the API and Codex.