Lawsuit Raises AI Data Risk
- Reports say a roughly $10 billion AI company was sued over its data-collection practices tied to major labs. - The lawsuit alleges problematic data practices used in training models linked with OpenAI and Anthropic partnerships. - Legal and copyright risks are becoming a material operational threat for AI startups and their investors. (x.com)
Mercor, a $10 billion artificial intelligence startup that supplies training work to major labs, is facing at least seven class-action lawsuits over how it collected and handled worker data. (timesofindia.indiatimes.com) The suits followed a security incident Mercor disclosed on March 31, 2026, tied to the open-source LiteLLM project. Mercor told TechCrunch it was “one of thousands of companies” affected and said outside forensics experts were investigating. (techcrunch.com) Mercor says it connects OpenAI, Anthropic and other labs with specialists such as doctors, lawyers and scientists to help train models, and the company said it was valued at $10 billion after a $350 million Series C led by Felicis in October 2025. (techcrunch.com, mercor.com) The complaints described in Wall Street Journal reporting go beyond the hack itself. Plaintiffs allege Mercor recorded job interviews, collected facial scans, captured screenshots of workers’ private computer screens, and shared some of that information with clients and partners. (timesofindia.indiatimes.com) That puts attention on a less visible layer of the artificial intelligence business: the contractors and data firms that generate feedback, examples and labels used to tune models after the base systems are built. Mercor’s own site says it powers “frontier research, RLHF data, and AI agent training at scale.” (mercor.com) The legal risk is not limited to one company. Scale AI, another major data-labeling firm, has also faced class-action litigation over contractor treatment tied to work used to train systems for companies including OpenAI, Google and Meta. (inc.com) Mercor has denied wrongdoing. The company said it “strongly dispute[s] the speculative claims” in the lawsuits, said it takes privacy seriously, and said it complies with relevant laws and regulations. (timesofindia.indiatimes.com) The breach has already had business fallout. TechCrunch reported on April 9 that Meta had paused Mercor contracts indefinitely, while OpenAI told Wired, as cited by TechCrunch, that it was investigating its exposure but had not paused or ended its contracts at that time. (techcrunch.com) Mercor’s rise showed how valuable human training data had become in the race to build better models. The lawsuits now put a price on the other side of that business: what happens when the data pipeline itself becomes a legal target. (mercor.com, techcrunch.com)
