Mercor faces class-action lawsuits after data breach exposes contractor information

A wave of class-action lawsuits has been filed against Mercor, a Silicon Valley-based artificial intelligence training startup valued at $10 billion, following a data breach that exposed sensitive information belonging to thousands of contractors. The legal actions, filed in recent weeks including one in Northern California on April 22, allege that the breach compromised highly personal data such as recorded job interviews, facial biometric information and screenshots from workers’ computers.

Mercor, which provides human feedback services used to train AI systems for major technology companies, confirmed that the breach involved a third-party partner and affected multiple organizations. The company’s business model centers on hiring contractors to evaluate and improve AI models for clients across the tech sector.

The lawsuits assert that the breach revealed not only personal data but also internal practices tied to how Mercor collects and uses contractor information. Plaintiffs claim the company monitored contractor computer activity, shared that data with clients and used recorded candidate interviews to train AI systems without sufficient disclosure. Legal filings also allege that applicant background checks and other vetting data were shared with partners in ways that may violate federal regulations, and that certain training materials may have included content owned by other entities.

Mercor has denied the allegations, stating that the claims are speculative and that it complies with all applicable laws and regulations. The company said it responded promptly to the breach and initiated a comprehensive investigation with third-party forensic experts. It also stated that it is communicating directly with affected stakeholders as more information becomes available.

The fallout has already impacted Mercor’s business relationships. Meta, a global technology company and one of Mercor’s clients, has paused its work with the startup while it reviews the situation.