Meta pauses AI training work with Mercor after data breach
04 Apr 2026

Meta has temporarily suspended all its work with data contracting firm Mercor, following a major security breach that affected the latter.

The indefinite suspension comes as other top AI labs also reconsider their partnerships with Mercor amid the ongoing investigation, WIRED reports.

The breach has raised concerns over potential exposure of sensitive training data used in AI products.

Mercor's role in AI training data generation
Data contracting

Mercor is one of the few companies that OpenAI, Anthropic, and other AI labs rely on for generating training data.

The firm employs a vast network of human contractors to create custom, proprietary datasets for these labs.

These datasets are usually kept under wraps as they play a crucial role in developing valuable AI models that power products like ChatGPT and Claude Code.

Potential impact on competitors and proprietary data
Data sensitivity

The data generated by Mercor is highly sensitive as it can give competitors, including other US and Chinese AI labs, insights into their training methods.

However, it remains unclear if the information leaked in Mercor's breach would significantly benefit a competitor.

OpenAI is currently investigating the incident to determine how its proprietary training data may have been compromised but has assured that the incident does not affect user data.

Mercor's confirmation of the security incident
Incident confirmation

Mercor confirmed the security incident in an email to its staff on March 31.

The company said it was a recent attack that affected its systems and thousands of other organizations globally.

Following this, contractors working on Meta projects have been barred from logging hours until (and if) the project resumes, leaving them in a state of uncertainty.

Contractors left in the dark about project status
Project reassessment

Contractors working on Meta projects were not given a clear reason for the pause.

A project lead on the Chordus initiative, a Meta-specific project to teach AI models how to use multiple internet sources to verify their responses, said that Mercor was currently reassessing the project scope.

This statement highlights the ongoing uncertainty surrounding this incident and its impact on future projects.