How to Stop AI Training on My Data: ChatGPT, Claude, Gemini and Copilot

Searching for how to stop AI training on my data? The short version: ChatGPT, Claude and Copilot each have a training switch in their settings, and Gemini has an activity switch with its own retention rules. Turning any of them off is not the same as deleting your chats, and some data is still kept either way. This guide uses only what each company says in its own help pages.

Quick answer table

ProviderDefault for personal accountsSetting nameWhat is still kept or used
ChatGPTData sharing for training is on for Free, Plus and ProImprove the model for everyoneExisting chats stay until you delete them; Temporary Chats may be kept up to 30 days for safety
ClaudeUsed if you allow it in Model Improvement; check your own Privacy Settings for your current choiceModel ImprovementSafety-flagged chats kept up to 2 years; thumbs up/down chats up to 5 years
GeminiKeep Activity is on for users 18 or overKeep ActivityConversations saved up to 72 hours even when off; a subset reviewed by humans and kept up to three years
CopilotUsed for training except excluded groups and users who opted outTraining on conversation activity; Training on voice conversationsOpt-out does not cover other product improvements, advertising, digital safety, security or compliance uses

The big takeaway: a training switch controls how future conversations are used. It does not erase what you already typed.

ChatGPT: one switch, plus Temporary Chat

OpenAI says data sharing for model training is enabled by default on personal Free, Plus and Pro accounts. To opt out, go to Profile > Settings > Data Controls and turn off Improve the model for everyone. New conversations are then not used to train OpenAI models.

Your existing chats are not deleted by this. OpenAI says you need to delete a chat or your account separately to remove content. The fact sources here do not give a retention period for deleted chats, so check OpenAI's help center if that matters to you.

Voice mode has its own separate data control, which can also be switched off. If you use voice, check it separately.

Temporary Chat

For a sensitive one-off question, Temporary Chat is built for that. These chats do not appear in history, do not create memories and are not used to improve models. They may still be retained for up to 30 days for safety purposes.

Work and business plans

OpenAI says that by default it does not use inputs and outputs from ChatGPT Business, Enterprise, Edu or the API to train its models.

Claude: Model Improvement and what it changes

For Claude Free, Pro and Max (including Claude Code on those plans), chats are used to improve Claude in three cases:

  • you allow it in the Model Improvement setting in Privacy Settings;
  • a conversation is flagged for safety review;
  • you explicitly opt in some other way.

If you allow Model Improvement, Anthropic may keep new or resumed chats in a de-identified format for up to 5 years in its training pipelines. The change applies only to new or resumed chats.

What stays the same when it is off

Safety-flagged conversations keep inputs and outputs for up to 2 years. Rating a response with thumbs up or down is another exception: the related conversation is stored for up to 5 years and may be used for training, de-linked from your user ID.

Deleting is a separate action. Deleted Claude conversations disappear from your chat history immediately and are deleted from Anthropic's back-end storage within 30 days. That 30-day figure belongs to deletion, not to the Model Improvement setting.

Incognito and work plans

Claude Incognito chats are not used to improve Claude even if Model Improvement is on. Anthropic's consumer-terms update does not apply to Claude for Work (Team and Enterprise), Claude Gov, Claude for Education, the API, Amazon Bedrock or Google Cloud Vertex AI.

Gemini: Keep Activity and the 72-hour window

Google says Gemini Keep Activity is on by default for users 18 or over. On a computer, go to gemini.google.com > Settings & help > Activity, click On, then choose Turn off or Turn off and delete activity.

Note what this setting is: Google's help page describes it as an activity setting. The facts used for this guide do not say it is a model-training switch, so read Google's own Gemini privacy page to see how it relates to training before you rely on it for that.

Here is what Google says about retention:

  • Even with Keep Activity off, Gemini saves conversations with your account for up to 72 hours to provide the service and process feedback.
  • With Keep Activity on, activity older than 18 months is auto-deleted by default. You can change this to 3 or 36 months, or turn auto-delete off.
  • A subset of Gemini chats is reviewed by human reviewers. Reviewed chats are retained for up to three years, disconnected from your account.

This page does not cover the default for users under 18. Check Google's help center for that.

Copilot: two switches and a list of exclusions

Microsoft says it uses data from Copilot for AI training except for certain categories of users and users who have opted out. Excluded from Copilot model training are:

  • signed-out users;
  • signed-in users under 18;
  • users signed in with an organizational Entra ID account;
  • Microsoft 365 Personal and Family subscribers;
  • users in Brazil, China (excluding Hong Kong), Israel, Nigeria, South Korea and Vietnam.

If you are not in one of those groups, look in Copilot's Privacy settings for two switches: Training on conversation activity and Training on voice conversations. Menu paths differ between app versions, so look for those names rather than a fixed route.

Microsoft says an opt-out is reflected throughout its systems within 30 days and applies to future conversation activity. It also says the opt-out does not exclude conversations from other product improvements, advertising, digital safety, security or compliance uses.

Separately, Microsoft stores Copilot conversations for 18 months by default. You can delete individual conversations or your whole history at any time.

Checklist

  1. Turn off the training switch where one exists: Improve the model for everyone in ChatGPT, Model Improvement in Claude, and the two training switches in Copilot.
  2. For Gemini, decide whether to turn off Keep Activity, and read Google's page on how it relates to training. Remember the 72-hour window.
  3. Use Temporary Chat (ChatGPT) or Incognito (Claude) for sensitive topics.
  4. Avoid thumbs up or down on sensitive Claude chats, since that can store the conversation for up to 5 years.
  5. Delete history separately. No training switch does that for you.
  6. On work or school accounts, check your organization's policy. Rules differ by plan.

FAQ

Does turning off training delete my old chats?

No. In ChatGPT, Claude, Gemini and Copilot the switch affects future conversations. To remove past content you have to delete chats or your history separately.

Are work or business AI plans used for training?

OpenAI says ChatGPT Business, Enterprise, Edu and the API are not used for training by default. Anthropic's consumer training setting does not apply to Claude for Work, Claude for Education or the API.

Is anything kept after I turn Gemini Keep Activity off?

Yes. Google keeps conversations with your account for up to 72 hours to provide the service and process feedback.

Is Copilot trained on chats from users in South Korea?

Microsoft lists users in South Korea among the groups excluded from Copilot model training, along with Brazil, China (excluding Hong Kong), Israel, Nigeria and Vietnam.

Sources

Facts last verified: 2026-10-12

Prices and plan limits come from the official pages listed above on the date shown and can change at any time. Check the vendor site before you subscribe. We have no partnership with the companies mentioned unless stated.

Comments

Post a Comment