NY judge orders OpenAI to provide ChatGPT conversations in a significant copyright case victory for newspapers.

A Manhattan judge has mandated that OpenAI disclose millions of anonymous interactions between its ChatGPT model and users, in a significant ongoing copyright infringement case. The ruling was detailed in a nine-page order made public on Wednesday, wherein Manhattan Magistrate Judge Ona Wang denied OpenAI’s motion to reconsider her previous directive from November, which required the company to provide 20 million logs of ChatGPT outputs to various media outlets, including the Daily News.

The media companies involved are pursuing an analysis of ChatGPT’s consumer logs to examine the model’s behavior concerning the reproduction of journalistic works. This ruling is part of a broader consolidated class-action lawsuit against OpenAI and its parent company, Microsoft, which was initiated in 2023. The lawsuit involves several prominent organizations, including The New York Times and outlets affiliated with Tribune Publishing and MediaNews Group, all alleging that the artificial intelligence technology improperly utilizes and distorts their copyrighted content. The suit also includes participation from the Authors Guild and multiple best-selling authors.

Frank Pine, Executive Editor of MediaNews Group and Tribune Publishing, expressed confidence in the legal proceedings, stating that OpenAI had previously underestimated the importance of providing evidence related to its business practices. Pine emphasized the need for OpenAI to compensate creators whose work has been used without authorization to develop and operate its technologies.

In contrast, a spokesperson for OpenAI referenced a blog post by executive Dane Stuckey, emphasizing the company’s commitment to user privacy. However, Judge Wang reiterated her stance that the users’ anonymity remains protected, noting that OpenAI had nearly completed the process of anonymizing the chat logs. She highlighted the multiple safeguards in place to mitigate privacy concerns inherent in the sensitive information being exchanged during the trial.

Wang identified the AI conversations as highly relevant to the media companies’ claims, asserting that they could contain either partial or complete reproductions of protected works. She declared that the request for production of the 20 million ChatGPT logs was proportionate to the case’s requirements, emphasizing that these logs constitute less than 0.05% of the total consumer output logs retained by OpenAI.

Following the completion of its deidentification process, OpenAI will have seven days to submit the requested data. The company has filed an appeal against Wang’s November ruling with Manhattan Federal Judge Sidney Stein, who oversees the case. Legal representatives for MediaNews Group and Tribune Publishing noted the importance of the judge’s findings, which implicated OpenAI in withholding critical evidence and raised questions regarding the company’s motivations for delaying the production of the logs.

This landmark decision sheds light on the intersection of artificial intelligence and copyright law, as well as the ongoing debate about how technology companies utilize content created by journalists and authors. As the case unfolds, it may have far-reaching implications for the future of AI and intellectual property rights.