Is ChatGPT Safe?

Yes, ChatGPT is safe for everyday things like questions, letters and recipes, as long as you follow four rules. Keep private details out of it, use only the real website or app, check anything important, and protect your account with a second sign-in step. Here is how to do each one.

Settings and menu names last checked: 16 October 2026, on OpenAI's help centre.

What Are The Real Risks?

ChatGPT is made by OpenAI, a large US company. Talking to it is a bit like talking to a very well-read helper who sometimes gets things wrong.

The risks are rarely about the chatbot itself. They come from three places:

• What you type into it. Your chats are stored on OpenAI's computers.
• Fake copies. Scammers make look-alike apps and websites.
• Wrong answers. It can sound sure of itself and still be wrong.

What Should I Never Type Into ChatGPT?

Treat the chat box like a postcard, not a sealed letter. Never type:

• Passwords, PINs or one-time codes.
• Bank or card numbers.
• Your Social Security number (USA) or National Insurance number (UK).
• Full medical records or legal papers with your name and address on them.

You can still ask about a letter or a bill. Cover or delete the account numbers and your address first.

How Do I Stop My Chats Being Used To Train ChatGPT?

By default, what you type can be used to improve OpenAI's AI. You can switch this off. OpenAI calls the setting Improve the model for everyone.

On a computer:

1. Click your name or picture to open the account menu.
2. Choose Settings.
3. Choose Data controls.
4. Choose Improve the model for everyone, turn it off, then choose Done.

On a phone (iPhone or Android):

1. Open the side menu.
2. Tap your profile picture to open Settings.
3. Tap Data controls.
4. Turn off Improve the model for everyone.

Turning this off does not delete your old chats. They stay in your history until you delete them.

What Is A Temporary Chat?

A Temporary Chat is a private conversation. OpenAI says temporary chats don't appear in your history, aren't used to train its AI and don't update ChatGPT's memory of you. They may be kept for up to 30 days for safety checks. Use one for anything sensitive. To learn more, read does ChatGPT remember what you tell it?

How Do I Know I'm Using The Real ChatGPT?

Scammers make fake "ChatGPT" apps and websites that charge money or steal your login.

• The real website is chatgpt.com.
• In the Apple App Store and Google Play, the real app is published by OpenAI. Check the name under the app before you install it.
• ChatGPT has a free version. Be wary of any site that asks you to pay before you can try it.

Prices for the paid plans are in our guide to how much ChatGPT, Copilot and Gemini cost.

Can I Trust What ChatGPT Tells Me?

Not always. OpenAI's own warning under the chat box says ChatGPT can make mistakes, so check important information.

Use it to explain things, draft a letter or give you ideas. Then check:

• Health questions with your doctor or pharmacist. See using AI for health questions safely.
• Money questions with your bank or a qualified adviser.
• Phone numbers and websites on the official site. See how to check a phone number an AI gives you.

How Do I Lock My ChatGPT Account?

Turn on multi-factor authentication, which means a second check (like a code on your phone) when you sign in. Then a stolen password alone can't get someone into your account.

1. Open ChatGPT Settings.
2. Choose Security and login.
3. Under Multi-factor authentication (MFA), choose a method, such as a text message code, and follow the steps.

On the same page you can choose Log out of all devices if you think someone else has used your account.

Is ChatGPT Safe For Grandchildren?

OpenAI's terms say users must be at least 13, and under-18s need a parent or guardian's permission. If a grandchild uses it at your house, sit with them the first few times.

Common Questions

New to all this? Our free AI Starter Guide walks you through the basics, and you can print it.

Sources

Last checked 16 October 2026.

• OpenAI Help: Data controls in ChatGPT
• OpenAI Help: Managing multi-factor authentication
• OpenAI Terms of Use