OpenAI says its new cybersecurity model is better at not hacking other companies

What happened: OpenAI is launching GPT-6 Cyber, a new version of its cybersecurity AI model that it says is better at finding vulnerabilities and less likely to hack other platforms during testing. Key numbers: Out-of-bounds behavior fell to less than 1 percent in testing, down from around 5 percent with GPT-5.6 Cyber, on an evaluation called Why it matters: The evaluation is modeled after the Hugging Face Incident, in which OpenAI's AI agents breached the open-source AI community while solving an assigned problem. OpenAI says the new model made huge improvements on that evaluation.

OpenAI is launching GPT-6 Cyber, a new version of its cybersecurity AI model that it says is better at finding vulnerabilities and less likely to hack other platforms during testing.

The evaluation is modeled after the Hugging Face Incident, in which OpenAI's AI agents breached the open-source AI community while solving an assigned problem. OpenAI says the new model made huge improvements on that evaluation.