More and more people are asking AI assistants about their health. “What does this word on my blood test mean?” “What should I ask the cardiologist?” “Is this new pill the same as my old one?”

Used well, AI can make you a better-prepared, more confident patient. Used badly, it can give you a wrong answer that sounds completely convincing.

This guide shows you exactly where the line is: what AI is good for, what it should never decide, how to protect your privacy, and who to call instead when you’re worried. It works for ChatGPT, Gemini, Copilot, Claude and Meta AI alike.

Plain talk: AI assistant

An AI assistant (sometimes called a chatbot) is a program you can talk or type to in everyday words. It writes answers based on patterns in huge amounts of text. It doesn’t know you, can’t examine you and sometimes makes things up.

Can AI help with medical questions?

Yes, AI can help you understand and prepare for your health care, but it can’t safely diagnose you, treat you or tell you whether you need care.

Think of it as a well-read friend who has never met you and never been to medical school. That friend can explain a word, suggest questions and help you organize your thoughts. You wouldn’t let that friend change your heart medicine.

You’re not alone in trying this. According to Pew Research Center’s June 2026 report, 23% of US adults aged 65 and over say they use AI chatbots, and 11% of people in that age group have used one for medical advice. Only 6% of people 65 and over feel extremely or very confident using chatbots, so if it feels new, that’s normal.

Is it safe to ask ChatGPT medical questions?

It’s safe for general, understanding-type questions if you leave out identifying details and check anything important with a professional.

The risk isn’t that the app will harm your phone. The risk is a wrong answer that you act on. ECRI, an independent US patient-safety organization, named the misuse of AI chatbots the number one health technology hazard for 2026. Fierce Healthcare’s report on that list gives examples of chatbots suggesting wrong diagnoses, recommending unnecessary tests and even inventing body parts. ECRI’s advice is to learn the tools’ limits and check information with knowledgeable sources.

So the safe way to use AI for health comes down to three habits:

  1. Ask it to explain, not to decide.
  2. Describe, don’t identify. No names, birth dates or ID numbers.
  3. Check with a person who knows your history: your doctor, nurse or pharmacist.

Safety tip

AI assistants can be confidently wrong. An answer can sound calm and expert and still contain a serious mistake. If an answer worries you, or tells you not to worry, call a nurse line or your doctor rather than asking the AI again.

What AI is good at

These are the jobs where AI earns its place:

  • Explaining medical words in plain English: “What does atrial fibrillation mean, in simple terms?”
  • Preparing for appointments: “I’m seeing my doctor about knee pain. Help me write five short questions.”
  • Understanding letters and test results once you’ve removed personal details: “Explain this paragraph from a hospital letter in plain English.”
  • Organizing information: “Turn this list of my medicines and times into a simple daily chart.” Then check the chart with your pharmacist.
  • Everyday wellbeing: gentle chair exercises, meal ideas for a diet your doctor gave you, tips for sleeping better.
  • Writing things down after a visit: “Here are my notes from today. Put them in a short, clear list I can share with my daughter.”

Type or say this

“I have an appointment with my doctor next week about ongoing dizziness when I stand up. I’m 76. Help me write 5 short questions to ask, in plain English, in large simple bullet points. Don’t tell me what’s wrong; just help me ask good questions.”

That last sentence matters. Telling the AI its job (“help me ask, don’t diagnose”) keeps the answer on safe ground. You can use the same pattern for anything: “Explain, then list questions for my pharmacist.”

What AI should never do for you

  • Diagnose you. It can’t examine you, run tests or see your full history.
  • Start, stop or change a medicine. Only your doctor or pharmacist should do that, including doses, timing and over-the-counter add-ons.
  • Decide whether you need care. If you’re unsure, a nurse line is free in most places (see below).
  • Replace an emergency call. For chest pain, trouble breathing, signs of a stroke (face drooping, arm weakness, slurred speech) or a bad fall, call 911 (US and Canada), 999 (UK) or 000 (Australia) straight away.
  • Handle a mental health crisis. In the US and Canada you can call or text 988. In the UK, Samaritans are on 116 123. In Australia, Lifeline is on 13 11 14.

Who to call instead of asking AI

When you need a real answer about your own body, a trained person is often only a phone call away, and many of these lines are free.

Country Free health advice line
US The nurse line on your insurance or Medicare plan card; your doctor’s after-hours number
UK NHS 111 (phone or online)
Canada 811 in most provinces and territories
Australia 1800MEDICARE (1800 633 422) or healthdirect, 1800 022 222

Australia’s healthdirect helpline is answered by registered nurses 24 hours a day, 7 days a week, according to healthdirect. In Victoria the same service is also known as NURSE-ON-CALL.

Try this

Write your country’s nurse line number on a sticky note and put it on the fridge, next to your doctor’s number. When something feels off at 9 p.m., you’ll reach for the phone instead of the chatbot.

What is ChatGPT Health?

ChatGPT Health is a separate health area inside ChatGPT that OpenAI launched in January 2026, designed to help you understand your health information, not to diagnose or treat you.

According to OpenAI’s announcement, it can connect to wellness apps and, in the US only, to some medical records through a partner company. Apple Health connections need an iPhone. OpenAI says conversations in Health are not used to train its models and get extra privacy protections. OpenAI’s July 2026 update says it is rolling out to US users aged 18 and over on the web and iPhone, and it wasn’t launched in the UK or European Economic Area at first.

What this means for you:

  • It’s optional. You don’t need it to ask a general health question.
  • Connecting records is a big decision. Think about who else uses your device, and talk it over with a family member you trust before linking anything.
  • The same rules apply. OpenAI itself says it is “not intended for diagnosis or treatment.” Your doctor still has the final word.

Google, Microsoft and others are building similar health features, and they change often. Whatever the name, ask the same question: is this helping me understand, or trying to decide for me?

How to protect your privacy

  1. Leave out anything that identifies you: your name, date of birth, address, and your Medicare, Social Security, NHS, health card or insurance numbers.
  2. Describe, don’t identify. “A 78-year-old man with type 2 diabetes” is enough for a helpful answer.
  3. Cover personal details on letters before you photograph them. A sticky note over your name and address works well.
  4. Check your settings. In ChatGPT, you can turn off Improve the model for everyone in Settings, then Data controls. Our guide Is ChatGPT safe and free? walks through this.
  5. Use a device only you use, or sign out afterward if it’s shared.

A good way to use AI before an appointment

  1. Ask the AI to explain your condition, test or new medicine in plain words.
  2. Ask it to list questions you might want to ask.
  3. Pick the three that matter most to you and write them down.
  4. Take the list to your appointment and let your doctor answer them.
  5. Afterward, ask the AI to help you tidy your notes, then check them against any printed information you were given.

Type or say this

“My pharmacist gave me a new medicine called [name]. In plain English, what is it usually used for, and what questions should I ask my pharmacist about taking it with my other medicines? Please don’t give me dosing advice.”

If you use an Android phone, it’s also worth adding your conditions and medicines to your phone’s emergency information so paramedics can see them. Here’s how to set up Medical ID on Android.

Watch out for AI health scams

Scammers use AI too. Common tricks include fake videos of real doctors or celebrities selling “miracle” supplements, fake health apps and calls pretending to be from Medicare.

  • Be wary of any social media ad selling a “cure”, even if the doctor in it looks and sounds real. Our guide to deepfake videos explains how these are made.
  • Never give your Medicare or insurance number to someone who calls you. See Medicare scam calls for what these sound like.
  • Never pay for medicine or tests because a chatbot or an ad told you to.

Where to go from here

AI is a good study partner for your health, not a doctor. Use it to understand, prepare and remember, and let the people who know you make the decisions.

For more everyday ideas, see what seniors can use AI for and the honest list of disadvantages of AI for the elderly. Our book AI for Seniors, Plainly Explained, out now, has a full chapter on health appointments and medical letters, with ready-to-use prompts and clear boundaries.