Meta suspends AI training with Mercor following data breach

Meta has stopped its association with Mercor and is looking into a recent data breach at the artificial intelligence (AI) training company, as a source close to the matter informed Business Insider.

Mercor, valued at $10 billion in a funding round last October, collaborates with major technology firms such as Meta to develop AI models aided by thousands of human contractors and specialists.

Wired first reported on Friday that Meta had put its work with the company on hold.

Mercor confirmed to Business Insider last Friday that it had suffered a security breach.

"Safeguarding our customers' and contractors' privacy and data is central to Mercor's mission. We recently detected that we were among the numerous companies affected by a supply chain attack linked to LiteLLM," Mercor said in a statement, mentioning the open-source initiative LiteLLM.

"Our security team acted swiftly to manage and resolve the issue," the company continued. "We are actively conducting a thorough investigation with assistance from top forensic experts."

Mercor has become an important player in the AI data pipeline, helping companies clean, label, and organise the vast datasets needed to build advanced systems.

Its client base includes some of the most prominent names in AI, though the full list of those impacted is still unknown.

What is clear, however, is that the breach may have revealed sensitive details such as data selection methods, labeling standards, and training strategies—assets that have taken years and significant investment to develop.