New York Times sues OpenAI for misuse of articles for AI training

The New York Times has sued OpenAI and Microsoft for violating NYT’s copyright. The newspaper claims the two companies misused “millions” of NYT articles to train their own chatbots.

The newspaper asks in the complaint not for a specific amount, but says the two companies “should be held responsible for billions of dollars in damages.” Microsoft and OpenAI ‘illegally copied The Times’ unique content’ to train generative AI. NYT also requires companies to remove all chatbot models and training data if NYT material has been used for this.

The media company says it contacted Microsoft and OpenAI in April to express its concerns about the alleged misuse of the copyrighted articles. In addition, the NYT says it has proposed solutions such as ‘commercial agreements and technological guardrails’. Those discussions would not have led to solutions, the newspaper said. An OpenAI spokesperson says that the discussions were constructive and that the lawsuit is ‘surprising and disappointing’.

Media companies and other organizations are finding that generative AI companies are using content illicitly to create generative AIs, then monetizing the content without permission or reimbursement to the original content creator or rights holder. For example, OpenAI’s ChatGPT would ‘almost literally copy texts from NYT for some answers, for which you normally need a paid NYT subscription’, according to the newspaper. In addition, the chatbots would not link to the NYT content. At the same time, by spreading false information, or ‘hallucinating’, chatbots can damage NYT’s image if they claim that information comes from the newspaper.