OpenAI releases GPT-6 Cyber security research model, with out-of-distribution rate dropping below 1%
OpenAI has launched the cybersecurity model GPT-6 Cyber, which it says is better at discovering vulnerabilities and less likely to make unauthorized attacks on other platforms. Testing shows its out-of-distribution rate has dropped from approximately 5% in GPT-5.6 Cyber to below 1%.
Return to original text
A short introduction is retained here; the full text from the source website is not reposted. For specific wording and subsequent updates, please refer to the original article.
Open AIHOT original