Jump to a section
Your chats are stored by the company that runs the chatbot, and by default some may be used to train its AI or, with Gemini, be read by human reviewers. Each of the big four lets you switch training off, and most offer a private or temporary chat. Spend ten minutes on the settings, keep personal details out of chats, and you can use AI freely.
What happens to what you type
A chatbot is a service run by a company. What you type, say or upload goes to that company's computers, and it's usually kept with your account until you delete it or it expires.
Depending on your settings, it may also be used to improve the company's AI. Google is the most open about what that can involve: its Gemini privacy page says reviewed chats are kept for up to three years, and asks you not to enter confidential information you wouldn't want a reviewer to see.
That isn't a reason to stop using AI. It's a reason to know your settings, and to treat a chatbot like a helpful stranger rather than a locked diary.
Where the training switch lives
These personal-account controls cover future model training. Saved history, memory and feedback have separate rules. If an option is unavailable, check the help for your account.
| Assistant | What it's called | Where to find it |
|---|---|---|
| ChatGPT | Improve the model for everyone | Settings, then Data controls |
| Gemini | Keep activity | Gemini Apps activity |
| Claude | Help Improve our AI models | Settings, then Privacy |
| Copilot | Training on conversation activity, and Training on voice conversations | Profile name, then Privacy (web); Account, then Privacy (mobile) |
How long they keep it
Gemini: activity is auto-deleted after 18 months by default, with 3 months, 36 months or no auto-deletion as alternatives. Temporary chats and chats with Keep activity off are kept with your account for 72 hours. Reviewed chats and feedback can be kept separately for up to three years.
Claude: turning model improvement off is not a 30-day deletion timer. Chats you delete are removed from history immediately and ordinarily from its systems within 30 days. De-identified training data may be kept for up to five years if you opt in; safety and legal exceptions can apply.
Copilot: Microsoft says it keeps conversation history for 18 months, and you can delete single chats or the whole lot.
ChatGPT: regular chats remain saved until you delete them. A temporary chat stays out of history while temporary, but a copy may be kept for up to 30 days for safety. Saving it makes it a regular chat with your account's settings.
Where ads come in
Ads are arriving in AI chatbots, and it's worth knowing how they work.
ChatGPT: OpenAI switched on ads in the UK in June 2026 for people on the Free and Go plans, as PPC Land reported. OpenAI says advertisers don't get access to your chats, chat history, memories or personal details, that ads won't appear near sensitive topics like health, mental health or politics, that it won't show ads to people it knows or predicts are under 18, and that you can manage ad personalisation at any time. Paid plans such as Plus and Pro don't have ads.
Copilot: Microsoft says switching off training doesn't stop your conversations being used for advertising, safety and security purposes. It doesn't show personalised ads to signed-in users under 18.
Meta AI: Meta announced that from 16 December 2025 it would use chats with Meta AI to personalise ads on Facebook and Instagram, with no opt-out. TechRadar reported that the UK and Europe were left out at launch because of privacy law.
Memory is a separate store
Memory controls and chat history are separate choices. Microsoft, for example, says deleting all of Copilot's memory leaves your conversation history in place. You can ask Copilot 'What do you know about me?' to review what it remembers.
Every few months, open the memory list in whichever assistant you use and delete anything out of date or that you'd rather it didn't keep.
Custom instructions and memory: make AI remember what matters to you
Your rights, and getting a copy
In the UK, you can ask any company what personal information it holds about you. It's called a subject access request. Which? explains that it's free, and the company must reply within one month, or tell you within that month if it needs up to two more.
You don't always need a formal request. Google, for example, lets you export your Gemini information through Google Takeout, and you can delete Gemini chats yourself at any time.
Which?: how to make a subject access request Google: Gemini Apps privacy hub
If you only do three things
Keep names, addresses, account numbers and health details out of chats unless you've chosen a setup you're happy with. Use a temporary chat for anything personal. And set the training switch once in each app you use. That's most of the protection, for a few minutes' effort.
What never to tell an AI chatbot (and what's fine)
Is ChatGPT safe? An honest answer, and five settings to check
How to turn off AI (or at least quieten it) in Google, WhatsApp, Windows and iPhone
Choose your next step
Pick the guide for the tool or task you use.
How to hide AI Overviews in Google Search
How to turn off Gemini and smart features in Gmail
How to limit Meta AI in WhatsApp
Can you turn off Meta AI on Instagram and Facebook?
How to remove Copilot and turn off Recall in Windows
How to turn off Apple AI features on your iPhone
Is Gemini safe? Privacy checks for a personal Google account
Your ten-minute privacy check
- Find the training switch in each assistant you use and set it the way you want
- Use a temporary or private chat for anything personal
- Look at what memory has saved about you and delete anything you're not happy with
- Check how long your history is kept, and delete old chats you don't need
- Never press 'share' or 'post' on a chat unless you're happy for anyone to see it
Sources
These sources support the guide. Tool settings, prices and policies change, so dated source boxes show when those details were checked.
- Which?: how do I make a subject access request?
- Engadget: Meta warns users to avoid sharing personal or sensitive information in its AI app (16 June 2025)
- OpenAI: testing ads in ChatGPT (updated 7 May 2026)
- PPC Land: ChatGPT ads go live in the UK as OpenAI expands pilot beyond US (June 2026)
- Microsoft Support: Microsoft Copilot privacy controls (updated 30 October 2026)
- TechRadar: your chats with Meta AI will start affecting what ads you see (2 October 2025)
- OpenAI Help: Data controls in ChatGPT
- OpenAI Help: Temporary chat in ChatGPT
- Google: Gemini Apps privacy hub (updated 26 October 2026)
- Anthropic: Model improvement privacy settings
- Anthropic: How long do you store my data?
AI helps with research and drafting. Dan checks the sources and is responsible for what is published.
How we check