Why the legal battle between the New York Times and OpenAI worries Europe
The US judge's decision to force Sam Altman's chatbot to preserve conversations is a privacy issue. The opinion of lawyer Giuseppe Vaciago (partner of 42 Law Firm)
3' min read
3' min read
ChatGPT will have to remember everything about us and that is a problem not only for us but also for our privacy. This time, however, the decision is not OpenAI's. Federal Judge Ona T. Wang ordered him to do so with a mandatory preservation order that prevents him from deleting any ChatGPT user conversations, even those deleted after the standard 30-day period, at the explicit request of the NYT in its copyright infringement lawsuit.
The order affects all output logs (chats and API responses), forcing OpenAI to store them indefinitely as long as the process is running.
The reason, as can be guessed, is related to the lawsuit that sees the New York Times against Sam Altman's company for copyright infringement: the accusation is that it has used millions of articles from the newspaper to train ChatGPT's language model without authorisation.
In order to find out whether the chatbot really did 'copy and paste' articles from the New York Times, the judge forced OpenAI to preserve all log data of ChatGPT conversations (prompts and outputs), fearing that the company was deleting potentially probative information. In the indictment, the US newspaper filed entire articles generated by the chatbot that appear to have been copied from the originals. The argument of Altman and his lawyers is that the technology, by generating text, never copies directly but reinterprets. Normally, OpenAI deletes users' chats on request or after a certain period (30 days, according to policy) for privacy reasons. Under the order of 13 May 2025, OpenAI is 'required to preserve and segregate all output log data that would otherwise be deleted, until further order of the court'. In other words, every ChatGPT conversation (input and response) will no longer have to be removed from OpenAI's servers, but stored separately as potential evidence, indefinitely (until otherwise decided). The decision, here in Europe, is raising quite a bit of controversy. Indeed, the issue is also relevant with respect to the European GDPR: although the court order is lawful under US law, it could conflict with EU law if OpenAI processes data of European users without valid GDPR grounds, e.g. without a clear legal basis or adequate measures to limit its use.
Access to this information will remain formally restricted to a very limited group of persons (OpenAI's internal lawyers undergoing security audits, and presumably the experts and lawyers of the litigants), as this data is only kept for court purposes. The judge's decision was taken in the specific context of a civil case and not as a general measure outside a concrete case.
