Hugging Face, the world's largest AI open-source community, has revealed a recent cyberattack on its servers by an AI agent. Initially, the security team attempted to utilize a U.S. commercial large model API to analyze over 17,000 logs related to the attack. However, the model's security mechanisms misjudged the requests, leading to a refusal of support. Subsequently, Hugging Face deployed China's GLM 5.2 open-source model on its infrastructure, which successfully conducted a forensic analysis of the logs. This process, which could have taken several days, was completed in just a few hours, highlighting the model's efficiency in handling the incident.