It is a fair question, and an important one: when you type something into ChatGPT, where does it go? The honest answer is “it depends on the tool and your settings” — but there are clear, simple rules that keep you safe without making AI useless. This guide walks through what actually happens to your data and what to never paste.
What actually happens to what you type
With many AI assistants, the conversations you have can, depending on your plan and settings, be used to help improve the underlying models. That does not mean a person is reading your chats, but it does mean sensitive information is best kept out unless you have turned that off. The safest assumption for anything private: treat the chat box like a semi-public space, not a locked drawer.
What you should never paste
- Passwords, PINs, and security answers. Never, in any AI tool.
- Financial details — full card numbers, bank account numbers, sort codes.
- Government IDs — passport, national insurance / social security, driver’s licence numbers.
- Other people’s private information — a customer’s details, a friend’s medical situation. It is not yours to share.
- Confidential work material your employer has not cleared for external tools.
Before pasting anything, ask: “Would I be comfortable if this appeared somewhere it should not?” If the answer is no, do not paste it — or strip out the sensitive parts first.
What's generally fine
Plenty is low-risk: general questions, drafting help, learning about a topic, brainstorming, editing text that contains no secrets, and working with information that is already public. You can also often anonymise — replace real names and numbers with placeholders like “[client]” — and get the same help without exposing anything.
Settings that protect you
Most major assistants now offer controls worth knowing:
- Turn off training on your chats. Many tools have a setting so your conversations are not used to improve the model. Find it and decide for yourself.
- Use temporary or incognito chats for anything you would rather not have saved to your history.
- Delete conversations you no longer need.
Use AI without oversharing
The free AI Prompt Builder helps you write effective prompts — no account, and nothing you enter is stored on our servers.
Try the AI Prompt Builder →A note on work and client data
This is where people slip up most. Pasting confidential company documents, unreleased plans, or customer data into a public AI tool can breach your employer’s policy or data-protection rules — even with good intentions. If your workplace has an AI policy, follow it; if it does not, ask before using client or company data. Anonymise where you can.
The bottom line
AI is safe to use for the vast majority of everyday tasks, as long as you keep genuinely sensitive information — passwords, financial and ID numbers, other people’s private data, confidential work material — out of it. Adjust your privacy settings, anonymise when in doubt, and you get all the benefit with very little risk.
Frequently asked questions
Not other users in the normal course of things. However, depending on your settings your chats may be used to help improve the model, and staff may review some conversations for safety. So keep genuinely sensitive information out, and use the privacy settings.
It helps — many tools stop using chats for training when history is off or a temporary chat is used. It is still wise not to paste passwords, financial details, or ID numbers regardless of the setting.
Only within your employer's rules. Confidential company or customer data should not go into a public AI tool without approval. Check your workplace's AI policy, and anonymise sensitive details where possible.
Anonymise it first — replace real names, numbers, and identifiers with placeholders. You usually get the same quality of help without exposing anything private.