Moonshot AI curbs Kimi K3.1 Pro hacking skills by scrubbing cyber data

Moonshot AI announced it removed cybersecurity-related data from training datasets of its new Kimi K3.1 Pro model to limit its offensive hacking capabilities, while preserving general programming skills. The move signals a proactive safety measure amid rising concerns over AI-enabled cyber threats, though the effectiveness and scope of the restriction remain unverified.

AI disclosure: Model-written summary of the source, not a verified fact. Check the sources. How Mosaic works