Meta freezes AI data work after breach puts training secrets at risk
Meta has suspended its collaboration with AI data startup Mercor after a supply chain attack exposed sensitive information about its training methodologies. The breach, which affected over 40,000 people, has triggered investigations and a class action lawsuit, sending a ripple of anxiety through the AI industry.
Meta has paused its work with Mercor, a San Francisco-based AI data company, after a cyberattack exposed sensitive information. The attack originated from a poisoned version of the LiteLLM open-source library, which was used to steal personal data and training methodologies. Mercor generates bespoke training datasets for big names in AI, including Meta, OpenAI, and Google. The breach has exposed approximately four terabytes of data. The incident has triggered investigations at OpenAI and Anthropic, and a class action lawsuit. Meta's pause is indefinite, and the incident has sent anxiety through the AI industry.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.