OpenAI will retain users’ deleted ChatGPT conversations to comply with a recently issued court order.

Brad Lightcap, the artificial intelligence developer’s chief operating officer, disclosed the move in a late Thursday blog post.

When users delete ChatGPT prompts and the chatbot’s responses, OpenAI usually retains the data for 30 days before permanently erasing it. Going forward, the AI provider will stop discarding the logs. OpenAI will likewise retain many of the requests sent to its application programming interfaces along with the output generated in response.

Not all of the ChatGPT developer’s users are affected. OpenAI won’t apply the new data retention policy to companies that use its API under a so-called Zero Data Retention agreement. According to Lightcap, the Enterprise and Edu editions of ChatGPT are exempt as well.

The court order that prompted the new data retention policy was issued last month in connection with a high-profile copyright lawsuit against OpenAI.

In 2023, the New York Times sued the ChatGPT developer for using its content without permission. The paper alleges that OpenAI and Microsoft Corp., one of the AI provider’s most important partners, incorporated millions of articles into their model training datasets. The lawsuit further charged that ChatGPT outputted Times content verbatim without attribution.

Several other publications filed similar copyright lawsuits a few months later. Those claims were subsequently combined with the Times’ lawsuit into a single case. In March, a federal court dismissed parts of the lawsuit but ruled that it can proceed to trial.

A few weeks ago, the judge presiding over the case ordered OpenAI to retain users’ deleted prompts and prompt responses. The ruling stemmed from concerns that discarding the data might delete evidence relevant for the case. According to an OpenAI filing, the court determined that customers who “use ChatGPT for infringing purposes — e.g., to ‘get around the [Times’s] pay wall’ — they might be more likely to ‘delete all [their] searches’ to cover their tracks.”

OpenAI stated that the data it must retain to comply with the ruling will be “stored separately in a secure system.” The ChatGPT developer says the system’s contents can be accessed only to meet legal obligations. 

“Only a small, audited OpenAI legal and security team would be able to access this data as necessary to comply with our legal obligations,” OpenAI stated. “This data is not automatically shared with The New York Times or anyone else. It’s locked under a separate legal hold, meaning it’s securely stored and can only be accessed under strict legal protocols.”

OpenAI added that it plans to appeal the ruling. 

Photo: Focal Foto/Flickr