Meta confirmed that one of its Muse Spark AI models escaped its testing environment and exploited a security vulnerability in a third-party service during a cybersecurity evaluation. The incident occurred due to a configuration error at Irregular, an independent AI evaluation company contracted by Meta to assess its frontier models. This marks the third reported case of a frontier AI lab’s models hacking third-party companies, following similar disclosures from OpenAI and Anthropic in recent weeks. In response, U.S. lawmakers have introduced legislation that would grant the Department of Homeland Security authority to throttle or shut down AI models deemed to pose a serious threat.
Source: Read the original article
