AI and your privacy: what happens to what you type

Start with the basics

Our guide on using AI safely lists what never to type into a chatbot. This guide explains what happens after you press send. It also covers the settings you can change.

What happens to what you type

When you send a message to a chatbot:

1. It goes to the company's servers, often outside Pakistan, where the model produces a reply.
2. It is usually stored, so you can see your past conversations. How long it is kept depends on the service and your settings.
3. It may be reviewed. Some companies let staff read samples of conversations to check quality and safety.
4. It may be used for training. Some services use conversations to improve future versions unless you switch this off.
5. Uploads go too. Files, photos and voice recordings you share are also sent and may be stored.

The rules can differ between free and paid versions. They can also differ between personal and work accounts. Read the privacy policy or help pages of the service you use.

Chat history

Most chatbots keep a list of your past conversations. This is convenient, but it means a record exists. Most chatbots let you delete single chats or your whole history in settings. Some offer a temporary chat mode that is not saved to your history.

Even after you delete a chat, the company may keep a copy for a limited time, for example for safety or legal reasons. The privacy policy usually explains this.

Memory

Some chatbots now have chat memory. They remember facts about you across conversations, such as your name, job, family or preferences. This can make answers more useful. It also means personal details are kept and used in later chats.

If the service has memory:

• Check what it has saved. Many services let you view and edit stored memories.
• Delete anything you do not want kept.
• Turn memory off if you prefer.

Training settings

Look in the app's settings for an option about using your data to improve the model. Turning it off can limit how your conversations are used. It does not always mean they are not stored.

Deleting your data

1. Download a copy first if you want to keep anything. Many services offer an export option.
2. Delete single chats you regret.
3. Clear memory if the service has it.
4. Check connected apps. If you linked the chatbot to your email, files or calendar, review or remove that access.
5. Delete your account if you no longer use the service. Most services explain how in their settings or help pages.

What never to share

The full list is in using AI safely. Two more things to think about here:

• Photos of children. Once uploaded, you lose control over where they go.
• Your voice. Long voice recordings of you or your family can help criminals make a voice clone.

Other places AI sees your data

• AI features inside phones, keyboards, email and messaging apps
• Browser extensions that offer to "summarise" or "improve" what you write
• Fake "AI" apps built to collect data. Download only from official app stores. See fake AI apps and investment scams

Each of these may send what you type or view to a company. Check their settings in the same way.

A monthly privacy check-up

Spend five minutes once a month:

1. Review and clear your chat history.
2. Check what memory has saved.
3. Confirm your training setting.
4. Remove AI apps and extensions you no longer use.

To understand the terms, see personal data in our glossary.