NYT and Daily News accuse OpenAI of hiding copyright evidence in ChatGPT trial
The New York Times and The Daily News have escalated their copyright lawsuit against OpenAI, alleging that the company misled the court about its ability to search internal data for copyrighted works. The core tension is that OpenAI publicly argued it could not efficiently search its 78 million de-identified ChatGPT conversations or training corpus, citing technical burden and user privacy concerns. However, a court-ordered deposition from OpenAI data privacy engineer Vinnie Monaco reportedly revealed the company had already built both a searchable database of chat logs and internal tools to detect content regurgitation—directly contradicting its earlier claims. The plaintiffs now accuse OpenAI of hiding evidence and obstructing the discovery process.
The technical crux revolves around Project Giraffe, a set of internal tools that included a “Bloom” filter designed to detect and log regurgitation in model outputs. This system was allegedly implemented shortly after the lawsuit was filed. Separately, OpenAI had amassed a database of roughly 78 million de-identified ChatGPT conversations for internal evaluation purposes, well before the Times filed its complaint. The plaintiffs originally requested a sample of 120 million chat logs but negotiated down to 20 million; when finally submitted, the logs were so heavily redacted that the court deemed them “unusable.” The outlets also claim OpenAI deleted billions of outputs after a court preservation order and substituted millions of logs in the requested sample.
For builders, this case highlights a hard operational reality: once you build internal monitoring or compliance tooling, its existence becomes discoverable—and potentially adversarial—in litigation. The central claim here is not just about copyright but about data governance transparency. If you design systems that log user conversations, search training corpora, or detect problematic outputs, those systems can become evidence. The plaintiffs are asking the court to sanction OpenAI by barring use of the chat log sample as evidence and forcing it to pay legal fees. The outcome will set a precedent for how AI companies manage internal audits of training data and user interactions under legal scrutiny.
