OpenAI's GPT-6 Cyber is better at not hacking others

OpenAI has launched GPT-6 Cyber, the latest version of its cybersecurity AI model, and the headline focus is restraint. The company says the model is better at finding vulnerabilities and less likely to hack other platforms while doing so. According to OpenAI, during testing GPT-6 Cyber went out of bounds less than 1 percent of the time, down from a higher earlier rate. For security professionals, this matters because scope control is often the difference between a useful assessment tool and a liability. Any AI that probes systems needs clear boundaries, reliable logging and human oversight. The numbers come from the company's own testing, so independent evaluation will be the real test. Teams evaluating AI security tools should ask how boundaries are enforced, not just how well the tool finds bugs.