What happens to Cursor Zero Retention policy now that OpenAI is forced to store everything, including API requests?
Hi @tmikaeld, OpenAI has to keep only ChatGPT data which is not related to Cursor as ChatGPT is the OpenAI Chat app and can not be used within Cursor.
Cursor uses for AI processing the OpenAI API which has no permanent storage and was not mentioned in the document you posted.
Hope that explains your concern.
Thanks! That makes it much better
The document mentions all output log data
that would otherwise be deleted, where does it specify that this does not apply to the OpenAI API?
That is the question because if this is enforced also on API nobody would want to use OpenAi.
So the real answer here is that we don’t know and this could currently already be enforced?
It’s still developing
Well, the sweeping court order that they have to comply with has been out there for two weeks, and that seems to include Zero Retention. So while the story might be developing, the court has ordered OpenAI to retain these logs. It’d be great that Cursor could provide a more reassuring response than giving roundabout answers like this.
Can or does Cursor add as default the parameter store=false
to its openai API calls for at least the Responses endpoint?
If ZDR is enabled, that Customer Content is discarded immediately after the inference completes, leaving only minimal system metadata. OpenAI never trains on API data even without ZDR; ZDR eliminates the 30-day log-retention window as well.
- Endpoints that can run in ZDR mode
• /v1/chat/completions (text-only; image inputs are excluded)
• /v1/completions
• /v1/embeddings
• /v1/audio/transcriptions, /v1/audio/translations, /v1/moderations – these already run with ZDR by default
• /v1/responses and new Responses-API tools also support ZDR whenstore=false
is used
While I do not know about their internal parameter settings,
Cursor Users can now update their privacy settings in Dashboard and would see:
Zero Data Retention
I didnt test all options but mine (Pro) shows this now.