Historic NYT v. OpenAI copyright battle heats up

A major copyright lawsuit filed by the New York Times against OpenAI and Microsoft in 2023 took a significant turn on Friday, with all three parties delivering their official arguments to a judge. The case, which could have far-reaching implications for the AI industry, hinges on whether the publishing giant's works were unlawfully scraped and used to train the AI models of its rivals.

In their motion for summary judgment, the Times claimed OpenAI and Microsoft had copied its content on a massive scale in order to create competing products. The tech giants, however, argued that existing copyright laws and legal precedents permit them to utilize publicly available works for training their large language models.

They contended that reported facts are not protected by copyright law and cited studies suggesting that ChatGPT's outputs do not constitute commercial substitutes for the Times' journalism. Microsoft echoed similar claims for its Copilot AI product.

OpenAI referenced two recent federal court decisions in California, Kadrey v. Meta and Bartz v. Anthropic, which found that using copyrighted works to train AI models amounted to fair use due to the transformative nature of the resulting AI outputs. The Times disagreed, arguing that those cases are not analogous as the plaintiffs did not demonstrate that AI outputs directly competed with their journalistic products, thereby displacing their market.

Additionally, the Times contends that the case should be assessed against established copyright precedents from higher courts, such as the Supreme Court and the Second Circuit, rather than the more recent lower court rulings cited by the defendants.

The legal battle has garnered global attention as it may set a crucial precedent for the future of AI. The Trump administration, which has expressed interest in AI infrastructure for national security and economic purposes, has also shown interest in the case. The Justice Department filed a brief in support of OpenAI, arguing that training large language models from millions of digital works provides a transformative public benefit that outweighs any competitive harm it may cause to the publishing industry.

District Judge Sidney Stein is anticipated to deliver his ruling in the coming months, determining whether the case, or specific parts of it, will proceed to trial next year. Axios, which has a licensing and technology agreement with OpenAI, has editorial independence from the lawsuit.