Short answer
General AI chatbots can be useful for explaining a diagnosis or preparing questions, but they should not be your only source for a health decision. The World Health Organization warns that these systems can produce false, inaccurate, biased or incomplete statements. Treat answers as a starting point and confirm anything that affects your care with a clinician or pharmacist.
Key takeaways
- WHO guidance on large multi-modal models warns they can produce false, inaccurate, biased or incomplete statements that could harm people making health decisions.
- Over-reliance is a named risk: a fluent answer can make users and professionals overlook errors.
- The FDA convened its Digital Health Advisory Committee on 6 November 2025 to discuss generative AI-enabled digital mental health devices, which shows regulators are still working out how to oversee them.
- A general chatbot is not a medical device you can assume has been tested for your situation.
- What you type about your health may be stored, so keep identifying details out.
What the risk actually is
Asking a chatbot about a rash, a lab result or a drug interaction feels like asking a knowledgeable friend. The difference is that a chatbot has no responsibility for the outcome and no view of your history. In January 2024 the World Health Organization published guidance on large multi-modal models, the family of systems behind popular chatbots. It warned that they can produce false, inaccurate, biased or incomplete statements, which could harm people who use that information to make health decisions.
WHO also flagged automation bias: users and health professionals may over-rely on these tools, overlooking errors or handing off difficult choices that should stay with a person. A smooth, confident answer is persuasive even when it is wrong. That is the same mechanism behind AI hallucinations and fake citations.
Where regulators stand
Regulators are still deciding how to treat AI that acts like a clinician. The US Food and Drug Administration lists a meeting of its Digital Health Advisory Committee on 6 November 2025 about generative AI-enabled digital mental health medical devices. The committee was asked to discuss benefits and health risks, ways to reduce those risks, and the evidence needed before and after a product reaches the market.
The practical point for readers is that a general chatbot sold as an assistant has not been through that kind of review for your situation. Do not assume that a friendly “wellness” or “therapy” label means a product has been tested or that a qualified person stands behind it.
Good uses and poor uses
Lower-risk uses are those where you stay in charge and can check the result:
- Explaining medical words in a discharge letter in plain language.
- Drafting a list of questions before an appointment.
- Summarising general information about a condition you have already been diagnosed with, then confirming details with your care team.
Poor uses are those where a wrong answer could hurt you:
- Deciding whether symptoms are serious enough to see someone.
- Changing a dose, stopping a prescribed medicine or combining supplements and drugs.
- Acting as a replacement for therapy or crisis support.
Your data is part of the picture
Health details are among the most sensitive things you can type. Before you paste anything, remove names, dates of birth, insurance or record numbers, and avoid uploading photos of documents. Our guide on what not to tell AI chatbots lists what to keep out, and chatbot privacy settings and data retention explains how to turn off history and training use.
Fake health tools and scams
Health is also a hook for fraud. Fake apps and extensions can pose as symptom checkers or AI doctors to collect data, which we cover in fake AI apps and malicious browser extensions. Be sceptical of any tool that asks for payment details before giving a “diagnosis”, or that pushes you toward a miracle product.
A simple rule
If the answer would change what you do with your body, get a second opinion from a pharmacist, nurse or doctor. If it is urgent, skip the chatbot and call emergency services.
What you can do
- 01Use a chatbot to learn terms, prepare questions for an appointment or understand a letter, not to decide whether to start, stop or change a medicine.
- 02For urgent symptoms such as chest pain, difficulty breathing or thoughts of self-harm, contact emergency services or a crisis line, not a chatbot.
- 03Check any specific claim, dose or interaction against a pharmacist, your prescriber or the official leaflet.
- 04Strip names, dates of birth, record numbers and photos of documents before pasting health information into a chatbot.
- 05Be wary of any chatbot or app that claims to be a licensed therapist or doctor, and look for who is accountable behind it.
- 06Switch off chat history or training use where the setting exists, and delete past conversations you no longer need.
Related reading: AI voice clone scams and the data breach tracker.
Frequently asked questions
Is it safe to ask an AI chatbot about my symptoms?
It can help you understand terms and decide what to ask a professional, but it can also be wrong or incomplete, as WHO warns. Do not use it to rule out something serious, and seek urgent care for severe or sudden symptoms.
Are AI therapy chatbots approved by regulators?
The FDA held an advisory committee meeting on 6 November 2025 on generative AI-enabled digital mental health medical devices, to discuss benefits, risks, risk mitigations, premarket evidence and postmarket monitoring. A chatbot that is sold as a general assistant has not been shown to be safe as a treatment.
Why do chatbots sound so sure when they are wrong?
They generate fluent text from patterns, so a wrong answer reads the same as a right one. WHO names automation bias, where people over-trust the output, as a risk. See our piece on AI hallucinations and fake citations for how this happens.
Does what I tell a chatbot about my health stay private?
Not necessarily. Many services keep conversations and some use them to improve models unless you change a setting. Treat health details as sensitive and share only what you need.
Sources
- WHO releases AI ethics and governance guidance for large multi-modal models, World Health Organization, 2024-01-18
- FDA Digital Health Advisory Committee, U.S. Food and Drug Administration, 2025-11-06
This article reflects public information as of 2026-11-06. AI tools and rules change quickly; we update articles when new facts are published. Spot an error? See our editorial and corrections policy.