AI Training Data Giant Mercor Is Reportedly Looking to Buy the Work You Did at Your Old Job
AI data giant Mercor is reportedly paying professionals for proprietary materials from past jobs, despite legal risks.
AI training data giant Mercor is reportedly attempting to purchase proprietary work materials directly from professionals, according to a Wall Street Journal investigation. The company has approached individuals in specialized fields like visual effects, offering payment for domain-specific assets such as '4D physics scenes with camera data, depth and motion/point tracking.' This strategy targets hard-to-acquire industry knowledge to train AI models, but immediately collides with intellectual property law. Most of this material is owned by the employers for whom it was created, and workers are typically bound by confidentiality agreements and contracts that forbid sharing such information.
Mercor claims in a statement that it 'does not buy intellectual property,' yet messages reviewed by the Journal show the company using the phrase 'looking to purchase' when contacting individuals about their prior work. This creates a legally dubious gray area where the outcome of such transactions would almost certainly involve IP. The controversy is compounded by a recent massive data breach at Mercor, where hackers allegedly stole 4TB of sensitive data including candidate profiles and employer information. This security failure underscores the significant risks for any professional considering selling protected materials, offering them little protection if the deal goes sideways.
- Mercor is offering payments to VFX artists and other pros for proprietary materials like '4D physics scenes' from past employers.
- The practice is legally fraught as the materials are typically owned by the original company and protected by NDAs and IP law.
- The news follows a major 4TB data breach at Mercor that exposed candidate profiles and employer data, highlighting security risks.
Why It Matters
This creates legal landmines for professionals and companies, accelerating the AI data gold rush while testing IP boundaries.