7 Things You Should Never Type Into an AI Chatbot

AI chatbots like ChatGPT can help you write a letter, plan a trip, or explain a confusing bill. They feel friendly, so it's easy to share more than you should. Here's a calm, simple guide to what you should never type into an AI chatbot, and what to type instead.

First, what happens to what you type?

An AI chatbot is a computer program you talk to in plain words. You type a question, and it types back an answer. ChatGPT, Google Gemini, Microsoft Copilot, and Claude are all chatbots.

Here's the part many people don't know. What you type doesn't vanish when you close the window. Most chatbots save your conversations in your account history. Some companies may also use conversations to train and improve their AI, unless you turn that setting off. Training means the company uses lots of examples to teach the AI to give better answers.

That doesn't make chatbots dangerous. It means you should treat a chatbot like a helpful stranger at the library. It's useful and polite, but you wouldn't hand it your wallet.

A good rule to keep in mind: if you wouldn't write it on a postcard, don't type it into a chatbot.

The 7 things to keep out of any AI chat

1. Passwords and PINs

Never type a password, a PIN, or a security question answer into a chatbot. Not even to ask, "Is this a strong password?" There's no need for the AI to see it.

Do this instead: Ask the chatbot to explain what makes a password strong. Then create your own, or let a password manager make one for you. A password manager is an app that creates and stores strong passwords, so you only remember one.

2. Bank, credit card, and account numbers

Your card number, bank account number, and routing number should never go into a chat. The same goes for photos of your cards or checks.

Do this instead: If you want help understanding a bank statement, describe the problem in general words. For example: "My statement shows a fee called a maintenance fee. What does that usually mean?"

3. Your Social Security number and ID numbers

Your Social Security number, driver's license number, passport number, and Medicare number are keys to your identity. Keep them out of every chat window.

Do this instead: If you're filling out a form and need help, ask about the form, not your details. Try: "What does this section of a passport renewal form usually ask for?"

4. Private medical details tied to your name

AI can explain medical words in plain English, and that can be helpful. But you don't need to share your full name, date of birth, or your complete medical records to get that help.

Do this instead: Ask general questions. "What does a high A1C number mean?" works fine without your name attached. And always confirm health advice with your doctor or pharmacist before acting on it. AI can be wrong, even when it sounds sure.

5. Other people's private information

It's tempting to paste in a family member's email or a friend's message and ask, "How should I reply?" But that person didn't agree to share their words or details with an AI company.

Do this instead: Summarize the situation without names. "My sister is upset that I missed her birthday. Help me write a kind apology." You'll get the same great help.

6. Work secrets and confidential documents

If you still work, volunteer on a board, or help with a family business, be careful. Contracts, client lists, and private plans don't belong in a public chatbot. Many employers have rules about this.

Do this instead: Check your workplace's policy first. When in doubt, describe the task in general terms, and leave out names, numbers, and private details.

7. Your exact home address and daily routine

Telling a chatbot "I live at 12 Oak Street and I'm away every Tuesday" isn't wise. You don't need your street address to get help planning a trip or a garden.

Do this instead: Use your city or region. "I live near Atlanta. What vegetables grow well here in the fall?" works great.

Why this matters, even if the company is trustworthy

You might wonder, "If I trust the company, why worry?" Here are 3 calm reasons.

1. Accounts can be broken into. If someone guesses or steals your chatbot password, they can read your saved chats. Anything private in there is now theirs.
2. Saved data can last a long time. Even after you delete a chat, companies may keep some data for a period of time under their own policies.
3. Scammers love personal details. The more facts about you that exist in more places, the easier it is for a scammer to sound convincing on the phone.

The fix is easy. Share less, and you have less to worry about.

How to use AI safely: 4 smart habits

These habits let you enjoy the help without the risk.

1. Use placeholders. Instead of real details, type stand ins like "[my name]" or "[account number]." The AI will still write your letter. Then you fill in the real details yourself, outside the chat.
2. Check your privacy settings. Most chatbots have a settings menu with privacy or data choices. In ChatGPT, look under Settings, then Data Controls, for a choice about whether your chats help improve the model. Other chatbots have similar options. Pick what you're comfortable with.
3. Try a temporary or private chat. Some chatbots offer a temporary chat mode that doesn't save the conversation to your history. Look for it when you're asking about something personal.
4. Delete chats you don't need. You can usually delete a single conversation or clear your whole history. A tidy history is a safer history.

Settings change often, so if you can't find an option, ask the chatbot itself: "Where do I find your privacy settings?" It will usually point you in the right direction.

Watch out for fake AI apps and AI scams

Here's a related warning, paired with the fix. Scammers make fake chatbot apps and websites that look like the real thing. Some ask you to "verify your identity" with a credit card or your Social Security number. A real chatbot won't need your Social Security number to chat with you.

The fix: Download chatbot apps only from your phone's official app store, and check that the company name matches. On a computer, type the website address yourself. For ChatGPT, that's chatgpt.com. Don't click chatbot links in emails, texts, or social media ads.

And remember: if a chatbot or anyone claiming to be one asks for money, gift cards, or passwords, stop. That's not help. That's a scam.

A quick example: the right way to ask

Let's say you want help writing a letter to dispute a charge on your credit card.

The risky way: "My name is Linda Harris, my card number is 4111 2222 3333 4444, and I live at 12 Oak Street. Write a dispute letter for a charge I do not recognize."

The safe way: "Write a short, polite letter to my credit card company disputing a charge I don't recognize. Leave blanks for my name, card number, and the charge details."

The second request gets you the same helpful letter. Your private details stay with you, where they belong.

This works for almost everything. Need a note to your doctor's office? Ask for a template with blanks. Want help with a letter to your landlord or homeowners association? Describe the issue, leave out the names and addresses, and fill them in yourself before you send it. You keep all the benefit and none of the risk. With a little practice, using placeholders becomes second nature.

Quick checklist

• No passwords, PINs, or security answers. Ever.
• No card, bank, Social Security, or ID numbers.
• No full name tied to private health details.
• No other people's private messages or details.
• Use placeholders like "[my name]" and fill them in later.
• Check your chatbot's privacy settings once.
• Delete chats you don't need.
• Download chatbot apps only from official app stores.

FAQ

Is it safe to use ChatGPT at all?

Yes, for most everyday tasks. Writing, explaining, planning, and brainstorming are great uses. The key is what you share. Keep private numbers and personal details out, and you can enjoy the help with confidence.

What if I already typed something private into a chatbot?

Don't panic. Delete that conversation from your history. If it was a password, change that password right away, and change it anywhere you reused it. If it was a card number, keep an eye on your statements and call your card company if you see anything strange.

Can the AI company's employees read my chats?

It depends on the company and your settings. Some companies say people may review some conversations to improve safety and quality. Check the privacy policy for the chatbot you use. The simplest protection is the same: don't type anything you'd mind a stranger seeing.

Want more simple, safe ways to use AI? Grab the free Digital Steady Starter Kit at /free-kit. You have got this. Stay steady out there.