A user hits a bug while Privacy Mode is on, switches the setting to Share Data, then sends support the request ID from that original conversation. What can support see, and what should the user do?

The principle — Privacy Mode is evaluated when a request is submitted, not when support later opens it. Each request is logged according to the mode in force at that moment, so the setting behaves like a stamp on individual requests rather than a switch governing an account's whole history.

Why the key is correct — Because the original request was submitted under Privacy Mode, its record holds only what that mode permits: which model ran, whether tool failures occurred without saying which, and backend failures unrelated to the prompt, code or agent actions. Switching to Share Data afterwards does not reach back and enrich it. The documented remedy is to enable Share Data, reproduce the problem, copy the new request ID, and send that identifier instead of the old one.

Why the others are wrong — Reading the setting as account-wide is the central error, since it grants no retrospective visibility however the report is worded. There is no propagation delay worth waiting out, because nothing further is being written for a request that has already finished. And the change deletes nothing: the earlier request remains on record, just too thin to diagnose agent behaviour from.

Remember this — A request ID is only as useful as the privacy mode it was created under, so reproduce first, then send the fresh one.

Sources — Cursor's Reporting a bug help page.