Using AI tools safely

Why safety matters with AI

AI tools such as chatbots, writing helpers and photo editors are now built into phones, email and office software. They save time, but they bring two kinds of risk. First, privacy: what you type or upload is sent to a company's computers and may be kept. Second, accuracy: the answers can be wrong while sounding certain (see how chatbots work).

Neither risk means you should avoid AI. It means using it with the same common sense you would use with any online service.

Protect your personal information

A good rule: do not type anything into a chatbot that you would not be happy to see on a postcard. Many services keep conversations for a time, and some may use them to improve their products unless you switch that off.

Keep these out of AI tools:

• Passwords, PINs and one-time codes. No genuine AI tool needs them.
• Bank card numbers, account numbers and ID document numbers.
• Medical details, legal matters or other people's private information, unless you are sure how the service handles it.
• Work documents, customer lists or anything confidential, unless your employer has approved that tool.

Tip: If you need help with a private letter, remove names, addresses and numbers first. Replace them with "[name]" or "[address]". The AI can still help with the wording.

Look in the app's settings or privacy section. Many let you turn off chat history, stop your chats being used for training, or delete past conversations. It takes two minutes and is worth doing.

Check before you trust

AI can make up facts, dates, prices, laws and even website addresses. This matters most for health, money, legal and safety questions.

1. Use AI to start your research, not finish it.
2. For anything important, confirm with an official source: your doctor, your bank, a government website or the company itself.
3. Ask the chatbot for its sources, then open them. If a link does not work or does not say what the AI claimed, do not rely on it.
4. Be extra careful with medicine doses, tax rules and legal deadlines.

Watch out for AI-themed scams

Criminals use the excitement around AI to trick people. Common tricks include:

• Fake AI apps in app stores or adverts that copy a famous name but steal data or charge hidden fees. Download only from the official website or a trusted app store, and check the developer name.
• "AI trading" or investment schemes promising guaranteed profits. Real investments never guarantee returns.
• Voice and video fakes of family members or public figures asking for money. See spotting deepfakes.
• Better-written scam messages. Criminals use AI to fix spelling, so good grammar no longer means a message is genuine.

Using AI fairly at work and school

Many workplaces, schools and universities now have rules about AI. Some allow it for brainstorming but not for final work; some require you to say when you used it. Check your organisation's policy before pasting in work material or handing in AI-assisted writing. If you use AI to help write something, read it carefully and make it your own. You are responsible for what you send.

If children in your home use AI tools, check the minimum age in the service's terms, sit with younger ones while they try it, and explain that the chatbot is a computer program, not a friend, however kind it sounds.

A quick safety checklist

• Use the official app or website, signed in with a strong password and 2-step verification.
• Check the privacy settings once, and switch off anything you are not comfortable with.
• Leave out passwords, card numbers and private details.
• Double-check facts that affect your health, money or safety.
• Be suspicious of any AI product that promises easy money.