New York Times accuses OpenAI of hiding evidence in copyright lawsuit, alleging deception over data access

The New York Times and The Daily News have escalated their copyright battle against OpenAI, accusing the AI company of concealing evidence and misleading the court about its ability to search training data and user chat logs for copyrighted content. The allegations, filed in a motion for sanctions, mark the latest flashpoint in a two-year-old lawsuit that could reshape the legal landscape for generative AI.

Allegations of Deception in Discovery Process

According to court filings, the plaintiffs claim that OpenAI falsely argued it could not search its training corpus or produce massive collections of ChatGPT conversations, citing technical burdens and user privacy concerns. However, an April deposition of OpenAI data privacy engineer Vinnie Monaco allegedly revealed that the company had already conducted internal searches and evaluations of its training data to identify copyrighted works before the lawsuit was filed.

Monaco’s testimony also reportedly disclosed that OpenAI had built a database of approximately 78 million de-identified ChatGPT conversations, used internally to assess potential infringement. Additionally, the company implemented a tool called the “Bloom” filter, part of a broader initiative named “Project Giraffe,” designed to detect and record instances of content regurgitation in outputs shortly after the lawsuit began.

Discovery Disputes and Redacted Data

The plaintiffs had initially requested a sample of 120 million chat logs, later negotiated down to 20 million. OpenAI submitted that sample last December, but the court reportedly found it so heavily redacted as to be “unusable.” The outlets further allege that OpenAI deleted billions of ChatGPT outputs after a court preservation order was in place, and substituted millions of logs within the requested sample, effectively obstructing the discovery process.

“If OpenAI genuinely believed that copying our clients’ journalism was fair and legal, it wouldn’t have hid the truth about having done it,” Ian B. Crosby, lead counsel for the plaintiffs, said in a statement. The plaintiffs are now asking the judge to sanction OpenAI, including barring the company from using the 20-million-log sample as evidence, and ordering it to pay legal fees incurred during the discovery dispute.

Why This Matters for the AI Industry

The case centers on whether OpenAI’s training of generative AI models on copyrighted news articles constitutes fair use or copyright infringement. If the court finds that OpenAI deliberately withheld evidence, it could undermine the company’s legal defenses and set a precedent for how AI firms handle discovery in intellectual property disputes. The outcome may also influence how other AI developers approach training data transparency and compliance with court orders.

OpenAI Responds, Denies Allegations

OpenAI spokesperson Drew Pusateri denied the claims, accusing the Times of trying to access private user conversations as its case weakens. “As the Times’ case weakens and they’ve been forced to drop claims against us, they’re persisting with their efforts to invade the privacy of people who have nothing to do with this case, including by making these blatantly false allegations,” Pusateri said. “We’ll continue defending our users’ privacy and the long-established principles of fair use.”

Conclusion

The escalating legal dispute underscores the growing tension between traditional media and AI developers over the use of copyrighted material in training data. As the court weighs the latest allegations, the case continues to test the boundaries of fair use in the age of generative AI, with potential ramifications for publishers, tech companies, and users alike.

FAQs

Q1: What is the core legal question in the New York Times vs. OpenAI lawsuit?
The central issue is whether OpenAI’s use of copyrighted news articles to train its AI models constitutes fair use or copyright infringement, and whether the company’s outputs reproduce that journalism without permission.

Q2: What new evidence did the plaintiffs claim to uncover?
The plaintiffs allege that OpenAI had already searched its training data for copyrighted works before the lawsuit, maintained a database of 78 million de-identified chat logs, and used a tool called the “Bloom” filter to track content regurgitation—all while arguing it could not do so during discovery.

Q3: What sanctions are the plaintiffs requesting from the court?
The New York Times and The Daily News are asking the judge to prevent OpenAI from using its 20-million-log sample as evidence, accept as fact that the logs would show major regurgitation, bar OpenAI from arguing otherwise, and order OpenAI to pay legal fees incurred during the discovery dispute.