How do Cami’s privacy settings work? First decide what you are managing this time: stop sending new content to third-party AI, cut down on proactive messages, correct a memory, or delete existing data. Each corresponds to a different entry point with a different effect.
Cami is a proactive AI companion for daily life; goals, memories, and reminders give conversations a follow-through. You should also know how that content is handled and where to adjust your choices. This article organizes common operations based on the full privacy policy, version of October 4, 2026. Menus depend on your app version. Processing that takes effect under the policy or requires separate consent follows the arrangements in the full policy.
First find the thing you want to manage
If you only want a feature to use less of one kind of data, look for that feature’s settings first. To stop further processing by third-party AI, manage AI data authorization. If you want existing records gone, you also need to request deletion or use the record-management entry points. Do not assume everything in your account is cleared just because you flipped one switch.
| What you want to do | Where to start | What else to check |
|---|---|---|
| Stop new sharing with third-party AI | Settings → Permissions → Privacy permissions → Manage AI data authorization | The result of the revocation and the affected features |
| Turn off personalized proactive messages | Settings → Push reminders → Personalized recommendations and proactive messages | Whether task reminders you set are still active |
| Edit or delete a memory | The memory management entry point | Whether the original chat, the memory, and feature records are each still stored |
| Export data or delete your account | Settings → Account & Security | The export scope, identity verification, and deletion details |
| Submit a personal information request | Settings → Help & feedback → Personal information request | The case number and the reply channel |
These entry points come from sections 3 and 7 of the policy. If your version does not offer the corresponding operation, you can submit a request through the email listed in the policy. There is no need to post private conversations in public comments.
Keep third-party AI authorization and device permissions separate
The system microphone permission allows audio recording; third-party AI authorization concerns handing content required by a feature to an AI provider for processing. Allowing the microphone is not blanket consent for other data uses, and agreeing to the general privacy policy, paying for membership, or having used AI before does not substitute for the separate AI authorization described in the policy.
When you review the disclosures under “Manage AI data authorization”, pay attention to the recipient, the purpose, and the scope of what gets sent. As the policy explains, after refusing or revoking, features that require third-party AI processing stop working. You can still view existing content, manage goals and tasks manually, and use local focus features that do not depend on third-party AI.
Revoking stops future sharing based on that authorization. Data already sent and requests already being processed cannot be recovered through this action, and history is not deleted automatically. To delete records as well, manage them separately or submit a deletion request. For the full explanation, see AI processing, memory, and sensitive information.
After deleting a chat, also check memories and feature records
The original chat, extracted memories, and miniapp records may be stored separately. For example, if you once chatted about “wanting to start reading this week”, a goal or memory formed afterwards should not be considered deleted just because the chat is gone. To fix goals and preferences, first look at what is actually stored in memory and in the relevant features.
I do not want to keep the old work schedule I mentioned before. I need to check where the original chat, the extracted memory, and the related tasks are stored, and delete what I no longer need.
This is an example of scoping what to manage. Saying this to Cami does not mean the data is already deleted. Use the corresponding entry points and check the results.
A miniapp’s AI reading or goal linking is managed in that miniapp’s settings. Turning off reading or linking is a different operation from deleting the original records. For how to look these up, see the Cami memory guide and the chat life-records guide.
Turning off proactive messages is not the same as canceling all reminders
If you want fewer suggestions and check-ins based on your personal traits, turn off “Personalized recommendations and proactive messages” at the entry point described in the policy. The policy states that this kind of proactive suggestion, care message, and report push stops, along with similar messages already queued but not yet sent. Messages already delivered are not recalled.
Conversations you start, task reminders you already set, agreed follow-ups, stage confirmations, operation and progress feedback, and necessary account notices still follow their own rules. If you no longer want a reading reminder, go back to that reminder’s own settings to check it, rather than only turning off personalized pushes.
System notification permissions are yet another layer. Before adjusting them, work out which kind of prompt you actually want fewer of, so you do not silence reminders for important tasks along the way. For task dates and repeat rules, see the Cami reminder settings guide.
Export, revocation, and account deletion: confirm each result
Exporting gets you the data described on that page and may require verifying your identity again. It does not include credentials such as passwords or tokens. After downloading to your own device, check that the file opens and contains what you need before considering anything else. Do not equate exporting with deleting.
As the policy explains, “Withdraw consent to the privacy policy” signs you out on the current device, stops the relevant services on it, and clears the consent record on that device. It does not delete your account, automatically delete historical data, or revoke account-level AI authorization. To stop third-party AI processing for your account on all devices, manage AI data authorization separately.
After account deletion, the account cannot be recovered, and cleanup of copies in various places does not finish the moment you tap. Information the law requires keeping, and backups that cannot yet be deleted immediately for technical reasons, are handled under the limits described in the policy with cleanup still scheduled. Uninstalling the app does not delete your account or server-side data. For the specific retention rules, see Information retention and deletion.
When the page cannot do it, how to submit a request
Submit through the “Personal information request” entry and keep the case number, or contact haohao@ixingtech.com. Say which kind of content you want to access, correct, or delete, and provide the account details needed for verification plus a way to reply. Do not send passwords or verification codes.
The policy commits to a response within fifteen business days of receiving a request; complex requests come with reasons, progress, and an estimated timeline. That response window does not mean deletion is complete across every medium. To learn the cleanup status, ask explicitly in a reply to the original request. For the full channels and rules, see Managing personal information.
FAQ: does using Cami mean agreeing to model training?
Are private chats approved for general model training by default just because AI is used?
The policy does not treat using AI features as that kind of blanket authorization. If such training were proposed, the purpose, scope, and participants would be explained separately, with the corresponding notice and consent steps. Do not read this rule as a reason to skip reading the specific authorization. HealthKit data is also subject to the dedicated purpose limits in section 2 of the policy.
After turning off a permission, do I still need to delete old records?
It depends on your goal. If you only want processing to stop, check that the revocation or switch-off completed. If you also want existing content deleted, manage the records separately or submit a deletion request. You can consider both together, but neither substitutes for the other.
Start by picking a feature you actually use from your daily routines, check what data and prompts it needs, then decide whether to keep, adjust, or stop it. Where anything is unclear, rely on the full privacy policy and the team’s actual replies.
Turn one idea into today’s action.
Get Cami ↗
