Short answer: the three major consumer chatbots handle privacy differently. In ChatGPT, turn off Improve the model for everyone to keep ordinary chat history while excluding new conversations from model training. In Gemini, turn off Keep Activity; that stops future chats appearing in Activity and generally prevents training, but it also removes the normal persistent history. In personal Microsoft Copilot, disable the conversation and voice training controls under Privacy. None of these switches automatically deletes old chats, memories, feedback records, or data retained for safety and security.
What you are changing—and what you are not
“Chat history,” “model training,” “memory,” and “deletion” are separate controls. A training opt-out usually affects future use of conversations to improve models; it does not necessarily hide chats already saved in your account. A history or activity switch controls whether conversations remain available in the account and may also control training, depending on the service. Memory controls personalization, while deletion removes selected chats or activity records under the provider’s retention rules.
As an Amazon Associate I earn from qualifying purchases.
History and training are different
ChatGPT lets you leave regular history enabled while switching off model improvement. Gemini combines its history/activity and training decision in one main setting. Copilot documents training controls separately from personalization. Read the result of each setting before assuming that “off” means no retention.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Memory is a separate privacy decision
A chatbot can remember preferences or facts for personalization even when a conversation is excluded from training. If you want to stop personalization, review the service’s memory or personalization controls separately and remove saved memories where the product provides that option.
#1 Best Overall
These instructions cover personal consumer accounts
Business, school, or government workspaces can impose administrator policies that override or hide consumer controls. Country, age or guardian settings, subscription plan, and app version can also change the labels. If a menu does not match the instructions below, open the provider’s current privacy or data-controls page from inside the signed-in app rather than assuming another account’s path applies.
“Not used for training” does not mean “not retained”
Providers may retain data for service delivery, abuse prevention, safety, security, legal compliance, or other product functions. ChatGPT says a Temporary Chat copy may be kept for up to 30 days for safety. Gemini says chats made with Keep Activity off, including Temporary Chats, remain associated with the account for 72 hours to provide the service and support safety. Copilot’s training opt-out does not exclude conversations from every other listed processing purpose.
ChatGPT: keep your history while opting out of model improvement
Turn off training for new conversations
- On the web: sign in, open the account menu, choose Settings, then Data controls.
- Switch off Improve the model for everyone and select Done if the confirmation appears.
- On iOS or Android: open the sidebar, tap your profile icon, choose Data controls, and switch off the same control.
OpenAI says this choice applies to the signed-in account across devices. New conversations can continue to appear in your chat history, but they will not be used to train OpenAI models. The setting does not delete, hide, or retroactively alter chats already stored in the account.
OpenAI’s documented wording is: “When Improve the model for everyone is off, your new conversations won’t be used to train OpenAI models. They can still appear in your chat history.”
Keep one conversation out of history with Temporary Chat
Use Temporary Chat when you want a particular conversation excluded from normal history and model improvement without changing the account-wide setting.
- Start a new chat.
- Select Temporary before sending your first message.
- Choose whether the temporary chat should be Personalized or Unpersonalized.
- Send your prompt only after checking that the Temporary label is active.
Temporary chats do not appear in history, do not create or update memories while temporary, and are not used to improve OpenAI models. A personalized Temporary Chat can still use existing memories, custom instructions, and plugins; choose the unpersonalized option if you do not want those applied. OpenAI may retain a copy for up to 30 days for safety. If you save a Temporary Chat, it becomes a regular chat and the account’s normal settings apply.
Delete existing chats or memories separately
Turning off Improve the model for everyone leaves saved conversations in place. Delete individual chats, clear history, or remove saved memories using the separate controls in ChatGPT settings when your goal is account cleanup rather than only a training opt-out. Deletion does not mean that information already incorporated into a model’s parameters can be instantly reconstructed or removed; the setting governs future use under OpenAI’s stated policy.
Gemini Apps: stop saving normal activity and training use together
Turn off Keep Activity
- Open Gemini Apps Activity while signed in to your Google account.
- Choose the control for Keep Activity.
- Turn it off and confirm the change.
- Return to the Activity view and verify that the status shows the new setting.
Google says future chats will not appear in Activity and will not be used to train its AI models unless you choose to send feedback. Unlike ChatGPT’s training switch, Keep Activity is also the normal history/activity control. Turning it off therefore removes the persistent, easily accessible Gemini activity record that many users expect. If you need an ongoing searchable history, understand that trade-off before confirming.
Understand Gemini’s 72-hour service retention
Keep Activity off does not create immediate zero retention. Google states that Gemini chats made while the setting is off, as well as Temporary Chats, are retained with the account for 72 hours to provide the service and support safety. That short-term retention is different from appearing in Activity or being used to train models.
Feedback creates an important exception
If you submit feedback about a Gemini response, Google may use the feedback and associated recent chat content. Google says specially trained teams may review that material, and reviewed feedback with associated conversations may be retained for up to three years after being disconnected from the account. Previously reviewed chats are not necessarily erased when you delete activity. Do not send feedback from a conversation that contains information you do not want considered under that process.
Keep Activity on but shorten automatic deletion
If persistent history is more important than a full training opt-out, Gemini lets you retain Activity while limiting how long it is stored. Google lists 18 months as the default auto-delete period and offers alternatives of 3 months, 36 months, or indefinite retention, plus manual deletion. These choices control Activity storage; they do not turn the setting into the same history-preserving training opt-out available in ChatGPT.
Use Temporary Chats for a single session
Gemini also offers Temporary Chats. They are not normal Activity entries, but Google still describes 72-hour retention for service and safety. Treat a Temporary Chat as a reduced-history mode, not as a promise that the content disappears immediately after you close the window.
Rank #4
Microsoft Copilot: disable conversation and voice training on a personal account
Microsoft’s documented controls are for personal consumer accounts. Its privacy page also warns that the older instructions do not apply to the updated Copilot app released from August 18, 2026. If your interface is newer, follow the linked privacy page presented in that app and check the labels before changing anything.
Web
- Open Copilot and select your profile icon.
- Select your profile name, then Privacy.
- Turn off Training on conversation activity.
- Turn off Training on voice conversations if you use voice features and do not want those conversations included in training.
Windows or macOS app
- Select the profile icon.
- Open Settings, then Privacy.
- Disable Training on conversation activity and, if applicable, Training on voice conversations.
Mobile app
- Open the menu and tap the profile icon.
- Choose Account, then Privacy.
- Switch off the conversation and voice training controls you do not want enabled.
Microsoft says opting out excludes future conversation activity from training the AI models. Its documentation also says this does not exclude conversations from other general product or system improvements, advertising, digital safety, security, or compliance purposes. The training choice is separate from personalization and memory: Microsoft says you may turn training off while leaving personalization enabled.
Microsoft’s documented wording is: “Opting out will exclude your future conversation activities from being used for training these AI models.”
Recommended Free Tools
How the three services compare
| Service | Can normal history remain while training is off? | What the main control covers | Documented short-term retention | Important exception or qualification |
|---|---|---|---|---|
| ChatGPT | Yes. Turn off Improve the model for everyone; chats can still appear in history. | Account-wide future model-training use; Temporary Chat handles one conversation. | Temporary Chat copy may be kept up to 30 days for safety. | Temporary Chat can use existing memories when Personalized; saving it converts it to a regular chat. |
| Gemini Apps | Not through the main opt-out. Keep Activity controls both Activity/history and training. | Future Activity saving and training use, unless feedback is submitted. | Keep Activity-off and Temporary Chats are retained for 72 hours for service and safety. | Feedback may be reviewed and associated material retained up to three years, disconnected from the account. |
| Microsoft Copilot | The documented training controls are separate from personalization; history behavior depends on the current app. | Future conversation and voice activity used to train AI models. | Not stated in the cited consumer control description. | Other product improvement, advertising, safety, security, and compliance processing can continue; app-version paths changed with the updated app from August 18, 2026. |
What to do after changing a privacy setting
- Check the status again. Reopen the same Data controls, Activity, or Privacy screen and confirm the switch is visibly off.
- Test with a harmless prompt. Send a non-sensitive message, then check whether it appears in history or Activity as the service documentation predicts.
- Delete old records separately. Remove chats, Activity entries, or memories that you no longer want stored. A training opt-out does not perform that cleanup.
- Review feedback history. Avoid submitting thumbs-up, thumbs-down, reports, or written feedback on sensitive conversations unless you accept the provider’s feedback-review terms.
- Check every device and account. ChatGPT’s account-level choice is documented as applying across devices, but a work profile, school account, or different signed-in account can have different policies.
- Review connected services. Data sent to a third-party plugin, connector, extension, or integrated service may be governed by that service’s own retention and training settings.
Common problems and the correct response
The named switch is missing
First confirm that you are signed in to a personal consumer account and that the app is updated. Workspace administrators can control or remove consumer options. Microsoft’s older Copilot privacy page specifically warns that its paths may not match the updated app introduced from August 18, 2026. Use the current in-app Privacy or Data controls page rather than relying on a screenshot from another version.
Best Value
The chat still appears after opting out
That behavior is expected in ChatGPT when only Improve the model for everyone is off: history can remain visible. In Gemini, Keep Activity off should prevent future chats from appearing in Activity, but short-term service retention can still occur. Seeing a chat is not, by itself, proof that it was used for model training.
You want history but not personalization
Keep ChatGPT’s normal history while turning off model improvement, then review memory and personalization separately. For a single ChatGPT session, choose an unpersonalized Temporary Chat. In Copilot, training and personalization are documented as separate choices. Gemini’s Keep Activity switch combines history and training, so it does not provide the same simple “keep all history, opt out of training” arrangement.
You already deleted the conversation
Deletion removes the account record according to the provider’s deletion process; it is not a time machine for model parameters or copies retained for safety, feedback review, legal compliance, or security. Google’s feedback process is especially important: reviewed feedback and associated conversations may remain for up to three years after disconnection from the account.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →You use voice input
On Copilot, conversation training and voice-conversation training have separate controls. Disable both if you want the documented opt-out to cover typed and voice activity. A typed-chat setting alone should not be assumed to cover voice data.
Which setting should you choose?
- Want ChatGPT history and no new model training: turn off Improve the model for everyone.
- Want one ChatGPT conversation absent from history: start a Temporary chat and choose Personalized or Unpersonalized deliberately.
- Want Gemini to stop building normal Activity history: turn off Keep Activity, accepting the 72-hour service-retention period and the feedback exception.
- Want Gemini history but less long-term storage: keep Activity on and choose a shorter auto-delete period such as 3 months, then delete sensitive entries manually.
- Want personal Copilot conversations excluded from model training: disable both conversation and voice training controls, then review personalization separately.
- Use a managed work or school account: consult the administrator’s policy; consumer instructions may not apply.
Limits of every opt-out
These controls address provider-side use of future consumer-app activity under each company’s published settings. They do not guarantee that information already used to train a model is removed from model weights, prevent all safety or security processing, or govern a separate connected service. They also do not change privacy settings in another product from the same company. For sensitive information, the safest practice is still not to enter it into a chatbot unless you understand the account, retention, feedback, and third-party terms that apply.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
