Meta has suspended its partnership with Mercor, a prominent data vendor serving AI companies, following a security incident that potentially compromised sensitive information about AI model training practices, according to WIRED AI. The breach has prompted multiple major AI laboratories to launch investigations into the extent of the exposure.
According to the report, the security incident at Mercor could have exposed key data detailing how various AI companies train their models, information that represents closely guarded industry secrets. The nature and scope of the compromised data remain under investigation as affected companies assess potential impacts.
The incident highlights growing security concerns in the AI industry’s supply chain, particularly regarding third-party vendors that handle sensitive technical information. As AI companies increasingly rely on external data providers and contractors, the Mercor breach underscores the risks associated with sharing proprietary training methodologies and datasets outside organizational boundaries.