What is the risk when an AI provider keeps your data?
When your team pastes client work into an AI tool, a copy can outlive the conversation. Three things can then happen to it.
A court can order it kept. That happened to OpenAI in NYT v. OpenAI, covered in the next section.
People can read it. On Azure, when content is flagged, "a sample of customer's prompts and completions may be selected for review" and stored for human reviewers, unless Microsoft approves you for modified abuse monitoring (Microsoft).
It can be stolen. Data written to storage can be taken from storage. Apple designed Private Cloud Compute so that data "must not be retained ... after the response is returned to the user", which leaves nothing at rest to take (Apple Security).
Can data an AI provider keeps be subpoenaed?
Yes. Data that exists can be subpoenaed, and a court can order it preserved. NYT v. OpenAI shows how. In that case, OpenAI was under a legal order to keep consumer ChatGPT and API content it would normally delete. OpenAI says: "Our obligations under the earlier order ended on September 26, 2025." In an October 2025 update, OpenAI said it still held limited user data from April to September 2025 (OpenAI).
The order covered ChatGPT Free, Plus, Pro and Team, and the OpenAI API without a zero data retention agreement. It did not cover ChatGPT Enterprise or ChatGPT Edu, or API customers using zero data retention endpoints. OpenAI explained why: "we never retain the prompts you send or the answers we return. Because it is not stored, this court order doesn't affect that data."
Faradex puts the same idea in two sentences: "You cannot subpoena what does not exist. You cannot preserve what was never stored." Faradex can be subpoenaed; it just has nothing to produce.
What does zero data retention not protect against?
- Saved chat history. Zero data retention is about processing. Proton erases data after Lumo answers, but stores saved history, encrypted. Kagi stores Assistant threads unless they are temporary.
- The companies behind the app. Duck.ai removes your identity, but its model providers can keep data for up to 30 days.
- Safety exceptions. OpenAI keeps images flagged as possible child sexual abuse material for review, and AWS may store and review flagged content to check for it.
- Metadata. Google still records token counts and latency under zero data retention.
- Features outside the scope. Zero data retention is set per feature or endpoint, so a feature it doesn't cover can keep your data. See How Zero Data Retention Works.