AI Trainer Mercor Offers to Pay People for Prior Work—Work Employers Might Own

Published
Score
13

Why it matters

Mercor, a San Francisco-based AI startup valued at $10 billion, is paying former employees of various industries—such as investment banks, consulting firms, and law firms—high hourly rates (up to $200) to provide expert knowledge for training AI models to automate those industries' workflows.[1][2][4][5][6] The core event involves Mercor's recent offer to compensate contractors for prior work experience, potentially including corporate knowledge that their ex-employers might claim ownership over, amid partnerships with AI labs seeking specialized data companies won't directly share.[2][4][6]

Key players include Mercor co-founder and CEO Brendan Foody (age 22), who champions the model arguing personal knowledge belongs to individuals, not firms; contractors (tens of thousands, often unemployed or supplementing jobs, paid $1.5M+ daily); and clients like OpenAI, Anthropic, Meta, and formerly Scale AI.[1][2][4][5][6] Mercor instructs against uploading proprietary documents but acknowledges risks of inadvertent sensitive data sharing given scale; job postings have sought access to production codebases, blurring lines on corporate espionage.[2][4]

Founded in 2023 by Foody and high school friends initially for AI-assisted hiring, Mercor pivoted to expert data labeling by 2024 after Scale AI requested specialized coders, scaling to $500M annualized revenue by late 2025 amid AI labs exhausting public datasets and spending $10B+ yearly on training data.[1][5][6] It shifted from low-skill overseas labor (e.g., Scale AI's model) to U.S.-based experts, with volatile contractor treatment like mass firings/hirings at lower pay.[1][2][3]

Newsworthy now due to the April 3, 2026 headline spotlighting ethical/IP tensions in a booming AI data market, as labs bypass reluctant firms by crowdsourcing ex-employees' insights—challenging work automation narratives, worker job security, and ownership of "knowledge in one's head" amid Mercor's billionaire-founder status.[1][2][4][5] Skepticism persists, with studies showing AI's real-world task failures (e.g., 70% by top models).[1]

Sources

mail Subscribe to Employment Law email updates

Primary sources. No fluff. Straight to your inbox.

Also on LawSnap