Is ChatGPT Safe? What Not to Share With AI

The short answer

For everyday tasks like drafting emails, planning meals, or explaining a topic, mainstream tools such as ChatGPT, Gemini, Claude, and Copilot are reasonable to use. The risks come mostly from three places: what you type in, trusting answers without checking, and fake apps or scams pretending to be AI tools. Each is easy to manage once you know about it.

What happens to what you type

When you chat with an AI tool, your messages are sent to the company's computers to produce a reply. Depending on the tool and your settings, conversations may be stored in your history, may be reviewed to keep the service safe, and in some cases may be used to improve future versions. Policies differ between companies and change over time, so the reliable rule is simple: don't type anything you wouldn't want a stranger at the company to read.

Treat the chat box like an email to a business, not like a diary.

Never share these

• Passwords, PINs, and security codes, including one-time codes sent by text.
• Government ID numbers such as Social Security, passport, or driver's license numbers.
• Full bank, card, or account numbers.
• Medical record numbers or insurance member IDs.
• Other people's private details: a friend's diagnosis, a coworker's personal situation, a grandchild's school and schedule.
• Confidential work information, unless your employer has approved the tool for that use.
• Your exact home address together with when you'll be away.

No legitimate AI tool needs any of these to help you.

How to get help with sensitive things anyway

You can still use AI for a medical letter, a bill, or a legal notice. Just remove or replace the identifying details first. Swap names for roles (“my doctor,” “the bank”) and replace numbers with placeholders like [account number].

Explain this letter in plain language. Tell me what it's about, whether I need to do anything, and by when. I've removed names and account numbers.

[paste the text with personal details removed]

I have an appointment with my [type of doctor] about [general symptom or condition]. Help me write a short list of clear questions to ask, in priority order. Remind me which things I should confirm with the doctor rather than rely on you for.

Five settings worth checking

Menus vary by app, but look in Settings (often under “Data controls” or “Privacy”) for:

1. Chat history: whether conversations are saved, and how to delete them.
2. Memory: some tools remember facts about you across chats. You can usually view, edit, or turn this off.
3. Training (improving the AI): an option about whether your chats can be used to improve the service.
4. Temporary or private chats: some tools offer a mode that isn't saved to your history.
5. Account security: turn on two-step verification if it's offered, and use a strong, unique password.

If you can't find a setting, ask the tool itself:

Where do I find the settings in this app for chat history, memory, and whether my conversations are used for training? Give me step-by-step directions for the [phone app / website].

Don't over-trust the answers

Safety isn't only about privacy. AI can state wrong information with total confidence, including invented facts, sources, and numbers. For anything involving your health, money, legal rights, or physical safety, use AI to understand and prepare, then confirm with a qualified professional or an official source. Never change a medication, sign a contract, or move money based only on an AI answer.

Use the real app, not a lookalike

Search results and app stores can include copycat “AI” apps. Before installing, check that the publisher matches the real company (for example, OpenAI for ChatGPT, Google for Gemini, Anthropic for Claude, Microsoft for Copilot). Be wary of apps that demand payment upfront, ask for unusual permissions, or show lots of ads. Bookmark the official website once you find it.

Watch for AI-powered scams

Criminals use AI too: to write convincing emails, and in some cases to imitate voices. The warning signs are familiar: urgency, secrecy, and requests for money, gift cards, or codes. If something feels off, stop and verify through a phone number or website you already trust. Read our full guide on how to spot AI scams.

A note for families

Many AI tools have minimum age requirements, often in the teen years, sometimes with parental permission. Check each tool's terms before letting a child use it. For younger kids, use it together, and talk about the same rules: no full names, schools, addresses, or photos of themselves.

Before I share anything, tell me what kinds of personal information I should avoid typing into an AI chat, and suggest how I could rephrase this request without it: [describe what you want help with].

One more habit: if you use a shared or family computer, sign out of AI tools when you're done, just as you would with email.

Your safety checklist

• Official app or website only.
• No passwords, codes, ID numbers, or full account numbers.
• Strip names and numbers before pasting documents.
• Check your history, memory, and training settings.
• Verify anything important with a real source.
• Pause on anything urgent that asks for money or codes.

Keep these handy with our printable AI cheat sheet. New to it all? Start with how to use ChatGPT for beginners.

General information only, not medical, legal, or financial advice. AI tools and their features change often; check the official site for current details.