Big AI companies rely on ingesting massive amounts of external content, including books, news articles, photographs, code, and websites, to train large language models. Court documents unsealed from a lawsuit involving The New York Times, OpenAI, and Microsoft reveal that Microsoft executives allegedly described the data intake as an astonishing theft of unprecedented proportions. An internal Microsoft policy document admitted that generative AI could disrupt the employment of data creators and destroy its own supply chain.
Judge Sidney H. Stein of the US District Court for the Southern District of New York has not yet issued a decision on the copyright case. OpenAI and Microsoft dispute the allegations using a fair use defense, claiming training models push forward public knowledge without acting as an unlawful economic market substitute. Meanwhile, sworn depositions indicate OpenAI models vacuumed up paywalled content without detection or removal efforts.
Code generation faces legal scrutiny
Code generation faces similar legal scrutiny. The US Ninth Circuit handed GitHub, Microsoft, and OpenAI a narrow win in the Doe v. GitHub lawsuit. The court ruled that generating new code without copyright management information does not equate to removing or altering existing copyright notices. Legal experts warn this decision leaves open source licenses vulnerable to contract litigation rather than traditional intellectual property enforcement.
Read nextOpenAI directs most research toward GPT 7 and beyondA federal antitrust lawsuit also targets Anthropic, OpenAI, SpaceXAI, and Google for allegedly coordinating to slow the development of advanced AI systems. Plaintiffs point to public statements from executives calling for pace controls on frontier AI development. Legal analysts note that antitrust claims require concrete evidence of harmful agreements beyond parallel public statements on safety.
The central debate focuses on whether AI developers can treat external work as free raw material while undercutting creator revenue. Courts must still resolve these overlapping intellectual property and market competition challenges.
