OpenAI will retain deleted ChatGPT conversations to comply with a recent court order.
This measure was announced by Brad Lightcap, the Chief Operating Officer of the AI developer, in a blog post on Thursday evening.
Typically, when users delete their prompts and the chatbot's responses in ChatGPT, OpenAI retains this data for 30 days before permanently erasing it. Going forward, however, the AI provider will cease discarding these logs. OpenAI will also preserve many requests sent to its Application Programming Interface (API) along with the generated response outputs.
Not all users of ChatGPT developers will be affected. OpenAI will not apply the new data retention policy to companies using its API under so-called zero data retention agreements. According to Lightcap, ChatGPT’s enterprise and educational versions are also unaffected.
The court order prompting the new data retention policy was issued last month during a high-profile copyright lawsuit against OpenAI.
In 2023, The New York Times sued the developers of ChatGPT, accusing them of using its content without permission. The newspaper claimed that OpenAI and one of its key partners, Microsoft, included millions of articles in their model training datasets. The lawsuit further alleged that ChatGPT reproduced The Times’ content without attribution.
Several months later, other publications filed similar copyright lawsuits. These lawsuits were subsequently consolidated with The Times’ case into a single proceeding. In March this year, a federal court dismissed parts of the litigation but ruled that the case could proceed to trial.
A few weeks ago, the judge presiding over the case ordered OpenAI to retain user-deleted prompts and responses. This ruling stemmed from concerns that discarding data might erase evidence relevant to the case. According to an OpenAI document, the court believed that "customers using ChatGPT for infringing purposes – such as 'bypassing [The Times'] paywall' – are more likely to 'delete all [their] searches' to cover their tracks."
OpenAI stated that the data required to be retained to comply with the ruling will be "stored separately in a secure system." The developers of ChatGPT noted that the contents of this system can only be accessed to meet legal obligations.
"Only a vetted OpenAI legal and security team will have access to this data if necessary to fulfill our legal obligations," OpenAI said. "This data will not be automatically shared with The New York Times or anyone else. It is locked in a separate legal hold, meaning it is securely stored and can only be accessed under strict legal agreements."
OpenAI added that it plans to appeal the ruling.
