Short answer: In ChatGPT, turn off "Improve the model for everyone" under Settings > Data Controls, or use a Temporary Chat. In Gemini, turn off Keep Activity (formerly Gemini Apps Activity). In Claude, turn off "Help improve Claude" under Settings > Privacy. All three settings stop future training. None of them stops storage. A ChatGPT Temporary Chat is kept for up to 30 days, a Gemini chat with Keep Activity off for up to 72 hours, and a Claude chat with training off for 30 days.
This guide walks through each setting as of November 2026, then shows what each service still retains afterwards and why "opt-out" is not the same as "zero retention".
ChatGPT: Temporary Chat and the training opt-out
Turn off training for all chats
- Click your profile icon (bottom left on the web, in the side menu in the mobile app).
- Open Settings, then Data Controls.
- Switch off Improve the model for everyone.
The setting syncs across devices when you are logged in. It applies to new conversations: anything you sent before switching it off may already have been used. Your chats still appear in your history.
You can also submit a "Do not train on my content" request through OpenAI's privacy portal, which records the choice at account level.
Use a Temporary Chat
- Open a new chat.
- Click Temporary at the top of the window. On the macOS app, click the model name at the top and turn on the Temporary Chat toggle.
A Temporary Chat:
- does not appear in your chat history,
- does not create new memories,
- is not used to train OpenAI's models,
- is deleted from OpenAI's systems after 30 days, and may be reviewed during that time only to check for abuse.
One catch: if you save a Temporary Chat, it becomes a regular chat and your normal settings apply to it.
What ChatGPT still keeps
| Situation | What happens |
|---|---|
| Training off, normal chat | Stays in your history until you delete it |
| Chat deleted | Removed from your account at once, permanently deleted within 30 days unless OpenAI must keep it for legal or security reasons |
| Temporary Chat | Kept up to 30 days, then deleted |
| Content already used for training | Deleting the chat does not remove it from a trained model |
The "legal reasons" exception is real. From May to September 2025, a court order in The New York Times v. OpenAI required OpenAI to preserve output logs that would otherwise have been deleted, including deleted and Temporary Chats. That obligation ended on 26 September 2025, but logs preserved before that date are still held, except those from users in the EEA, Switzerland and the UK. More on that in Is ChatGPT safe?
ChatGPT Business, Enterprise and Edu workspaces, and the API, are not used for training by default.
Gemini: turn off Keep Activity
For users aged 18 and over, Keep Activity is on by default. Google renamed the setting from "Gemini Apps Activity".
On a computer
- Go to gemini.google.com and sign in.
- Click Settings & help, then Activity (or go to myactivity.google.com/product/gemini).
- Click On at the top and choose Turn off, or Turn off and delete activity to also clear older chats.
On Android
- Tap the menu, then your profile picture.
- Tap Gemini Apps Activity (labelled Keep Activity in newer versions).
- Tap On and choose Turn off or Turn off and delete activity.
With the setting off, future chats are not used to improve Google's AI models unless you send feedback. On a personal account, your chat history is no longer saved either.
What Gemini still keeps
- Up to 72 hours. Even with Keep Activity off, Google keeps conversations with your account for up to 72 hours to provide the service and process feedback. No setting shortens this. Temporary Chats are kept for the same 72 hours.
- Up to 3 years for reviewed chats. Chats that human reviewers looked at (plus related data such as language, device type and location info) are not deleted when you delete your activity. Google keeps them for up to three years, disconnected from your account.
- With Keep Activity on: chats auto-delete after 18 months by default. You can change that to 3 or 36 months, or turn auto-delete off.
On a Google Workspace work or school account, Gemini runs under your organisation's Workspace agreement. Google states that those chats and uploaded files are not reviewed by humans or used to train generative AI models outside your domain without permission. There, your administrator controls the activity setting.
Claude: turn off "Help improve Claude"
Steps
- Click your name in the bottom left corner and open Settings.
- Choose Privacy.
- Switch off Help improve Claude.
The choice applies to new and resumed chats and coding sessions. When Anthropic introduced the setting in its consumer terms update of August 2025, existing users had to choose by 8 October 2025, and the toggle in that pop-up started switched on. If you clicked "Accept" without looking, training is probably on.
Incognito chats
Incognito chats (the ghost icon) are not used to train Claude even if model improvement is switched on, and they are not saved to your history or to Claude's memory.
What Claude still keeps
- Training off: conversations are retained for 30 days.
- Training on: retention extends to up to five years, in de-identified form.
- Deleted chats: removed from Anthropic's back-end systems within 30 days.
- Feedback: if you rate a response or send feedback, the related conversation can be kept for up to five years, whatever your setting.
- Safety flags: conversations flagged for a usage policy violation fall under separate, longer retention.
The setting covers Claude Free, Pro and Max, including Claude Code on those plans. Claude Team, Claude Enterprise and the API run under Anthropic's commercial terms and are not used for training by default. The full breakdown by plan is in Claude privacy explained.
Comparison: what each service keeps after you opt out
| Service | Where to opt out | Retention after opting out | Temporary / incognito mode |
|---|---|---|---|
| ChatGPT (Free, Plus, Pro) | Settings > Data Controls > Improve the model for everyone | In history until deleted, then up to 30 days | Temporary Chat: up to 30 days |
| Gemini (personal account) | Settings & help > Activity > Turn off | Up to 72 hours; reviewed chats up to 3 years | Temporary Chat: up to 72 hours |
| Claude (Free, Pro, Max) | Settings > Privacy > Help improve Claude | 30 days (up to 5 years with training on) | Incognito: not used for training |
Based on each provider's published help and privacy pages as of November 2026. Setting names and menus change; check the current version for your plan.
Why opting out is not the same as zero retention
An opt-out controls one thing: whether your conversation becomes training material for a future model. It does not control whether the conversation is stored. While it sits on the provider's servers, it can be checked by abuse-detection systems, opened by reviewers in some cases, and, as the 2025 court order showed, preserved in litigation.
For a recipe or a travel plan, that hardly matters. For a client's contract, a patient's history, an unreleased product or a salary negotiation, the question is not only "will this be trained on?" but also "where will it sit, for how long, and who could be made to hand it over?" Our privacy comparison of ChatGPT, Claude and Gemini covers that side in more depth.
What zero data retention means
Zero data retention (ZDR) is an arrangement in which the model provider does not store your inputs and outputs. The request is processed long enough to generate the answer and then discarded: no 30-day window, no 72-hour buffer, no review queue. You cannot switch it on in a consumer settings menu. It is agreed between businesses in a contract. See what zero data retention is for the details.
How Wysor handles this
Wysor is an AI workspace connected to the model providers under zero data retention, so there is no opt-out to remember.
- Zero data retention with the providers behind Claude, GPT, Gemini, Grok, Mistral, Perplexity and the open-source models in Wysor.
- No training on your data, on any plan.
- Hosted in the EU, on AWS in Frankfurt.
- Data processing agreement included.
- Your choice of model, switchable in the same chat.
There is a free plan, Plus at €19.99 per month and Premium at €29.99 per user per month. If you are comparing privacy-focused tools more broadly, see private AI chat: assistants that don't train on your chats.
Frequently asked questions
Does ChatGPT Temporary Chat delete my conversation? Yes, but not immediately. A Temporary Chat never appears in your history and is not used for training, but OpenAI keeps a copy for up to 30 days for safety purposes before deleting it.
If I turn off "Improve the model for everyone", is my chat history deleted? No. Your conversations stay in your history. Training stops for new chats; to remove old ones, delete them, and they are then permanently deleted within 30 days.
Can I stop Gemini from keeping my chats for 72 hours? No. Even with Keep Activity off, Google keeps conversations for up to 72 hours to run the service and process feedback. There is no setting that shortens this.
Is Claude trained on my chats by default? On Claude Free, Pro and Max, it depends on what you chose in the 2025 consumer terms pop-up, where the training toggle started switched on. Check Settings > Privacy > Help improve Claude. Claude Team, Enterprise and the API are not used for training by default.
Do paid plans like ChatGPT Plus or Claude Pro protect me from training? Not automatically. Individual paid plans follow the same consumer rules as the free versions, so you need to check the setting yourself. Business plans (ChatGPT Business and Enterprise, Claude Team and Enterprise, Google Workspace) are excluded from training by default. More in Is the free version of ChatGPT or Claude enough?
Is it safe to paste confidential information if training is off? Training off reduces one risk, not all of them: the conversation is still stored for a period and handled under consumer terms without a data processing agreement. For client, patient or company data, look for zero retention, no training in the contract, and a DPA.
