Moonshot limits hacking ability in new Kimi model

Published Oct 26, 03:23 AM UTC

Moonshot has curbed the hacking ability of its new Kimi model, according to a post that frames the change against growing fears over AI-powered cyberattacks. The item presents the restriction as a safety-oriented adjustment to the model's capabilities, though it does not specify the technical methods used, the exact capabilities removed, or the timing of the change. It also does not name any external regulator, research group, or incident driving the concern. The post carries hashtags related to AI safety, AI regulation, AI governance, cybersecurity, and cyberattack, indicating that the development is being positioned within broader policy and security debates. No independent confirmation, company statement, or quantitative benchmark is included in the source text. The linked URL is not reproduced here, so no additional details from the destination page are available for verification or context.