Big AI firms accused of stealing creators’ work to train language models

Recent legal documents reveal Microsoft staff labeling the mass ingestion of online material for AI training as possibly the "largest theft of labor in human history," while OpenAI admitted its systems routinely scrape pay-walled content without detection. A Ninth Circuit ruling limited a DMCA claim against GitHub Copilot, but did not clear the way for unrestricted training on open-source code. Separate antitrust filings allege that Anthropic, OpenAI, SpaceXAI and Google coordinated to pace frontier AI development, which could restrict competition.

Plaintiffs in the copyright case argue that AI firms profit from creators’ work without compensation, and OpenAI’s co-founder responded dismissively to concerns about bypassing firewalls. The combined disputes highlight growing tension between AI developers, content creators, open-source communities and regulators.