Is AI safe to use? Privacy, scams and your settings
Mostly, yes, if you know what to keep to yourself. Here’s what happens to your chats, the settings worth changing and the AI scams to watch for.
Video transcript
Is it safe to type your life into an AI chat? Mostly, yes, if you follow a few simple rules. Here’s what happens to your chats, and how to stay safe.
Whatever you type goes to the company that runs the assistant. It may be kept, and used to improve the AI, unless you switch that off in your privacy settings.
So some things should never go in: passwords, bank card details, other people’s personal information, and confidential work. If you’d hate to see it leaked, leave it out.
Protect your account like your email. Use a strong password you don’t use anywhere else, turn on two-factor authentication, and use a temporary chat for anything you’d rather not keep.
Criminals use AI too. They can fake a video of a celebrity, clone a loved one’s voice from a short clip, write flawless scam emails, and build fake AI apps.
So agree a code word with your family. If someone calls asking for money, ask for the word, or hang up and call them back on a number you know.
If you’re targeted, report it to Report Fraud, or to Police Scotland if you live in Scotland. Read the full guide below for the settings worth checking today.
In 30 seconds
- Your chats may be stored and used to train the AI, so check your privacy settings.
- Never share passwords, card details, other people’s information or confidential work.
- AI makes scams more convincing: agree a family code word and call back on numbers you know.
Typing into an AI chat can feel as private as scribbling in a notebook. It isn’t quite: your words go to the company that runs it, and may stay there. That’s no reason to avoid AI, just a reason to use it wisely.
What happens to your chats
When you use an Jargon busterAI assistant: A chat app such as ChatGPT, Claude or Gemini that answers questions and does tasks you describe in plain language. such as ChatGPT, Claude or Gemini, your messages are sent to the company’s computers. Here’s what can happen next.
- They’re stored. Your history stays in your account until you delete it, and companies may keep copies for a while after that.
- They may help train the AI. Many assistants can use your chats to improve their models unless you switch this off. Look in
Settings, under something likeData controls,PrivacyorActivity. - Some may be read by staff, for example when checking for misuse.
- Temporary chats are the exception. Many assistants offer a temporary or incognito chat that isn’t saved to your history or used for training.
What to keep out
A simple rule: if you’d hate to see it leaked, leave it out. You can usually get the same help by removing the personal bits first.
| Fine to share | Keep it out |
|---|---|
| General questions, such as “Is this text a scam?” | Passwords, PINs and security codes |
| A letter with names and account numbers removed | Bank card and account details |
| Your own writing, to tidy up or shorten | Other people’s personal details, such as a friend’s health |
| A work problem described in general terms | Confidential work files or customer data |
Using AI for your job? Your employer may have its own rules and tools: see how to use AI at work safely.
Keep your account safe
Your chat history can reveal a lot about you, so protect it like your email. Use a strong password you don’t use anywhere else, and turn on Jargon busterTwo-factor authentication: A second check when you log in, such as a code sent to your phone, so a stolen password isn’t enough., either in the assistant’s settings or on the Google, Apple or Microsoft account you sign in with.
Try it yourself 5 minutes
- Open your assistant’s
Settingsand find the data or privacy section. - Choose whether it may use your chats to improve its AI.
- Find the temporary or incognito chat option, ready for anything you’d rather not keep.
- Check that two-factor authentication is on for the account you sign in with.
AI-powered scams to know about
Criminals use AI too. It makes Jargon busterPhishing: A scam message pretending to be from someone you trust, to trick you into sharing details or paying. messages more convincing, and it can fake faces and voices. The best defences are simple, and they work however good the fake is.
- Deepfake videosA Jargon busterDeepfake: A fake video, image or voice recording made with AI to look or sound like a real person. can show a famous face backing an investment. Treat any “get rich quick” video as a scam, however real it looks.
- Cloned voicesJargon busterVoice cloning: Using AI to copy someone’s voice from a short recording, so it can say things they never said. can copy a voice from a short clip. If “your son” calls needing money, hang up and ring his usual number.
- A family code wordAgree a word only your family knows. If a caller asking for help can’t say it, it isn’t them.
- Flawless phishingAI can write scam messages with perfect spelling. Judge the request instead: urgency, links and demands for money are the warning signs.
- Fake AI appsDownload assistants only from official app stores, and check the developer is the real company before you install.
I’ve received the message below and I’m not sure it’s genuine. I’ve removed my personal details. [paste the message] What are the warning signs that it might be a scam? How can I check whether it’s real without using any links or phone numbers in the message?
If you’re targeted, report it to Report Fraud, the national service that replaced Action Fraud, or to Police Scotland on 101 if you live in Scotland, and forward scam texts to 7726. If a company mishandles your personal data, you can complain to the ICO, the UK’s data protection regulator. And before using AI with your finances, read AI and your money.
Check yourself
3 quick questions nothing is savedTools in this guide
Spotted a mistake? Tell us and an editor will check it.
