News Summary

Meta has indefinitely paused its collaboration with Mercor, a data contracting firm, following a security breach that exposed sensitive AI industry secrets. Mercor, which provides training data for AI labs like OpenAI and Anthropic, confirmed the incident affected its systems and thousands of other organizations globally. This breach has prompted other AI labs to reassess their partnerships with Mercor. The incident did not affect OpenAI user data, but OpenAI is investigating the exposure of its proprietary training data. The breach, linked to a hacker group known as TeamPCP, also compromised updates of the AI API tool LiteLLM, affecting numerous companies. The breach highlights the risks associated with the secrecy and security of AI training data.