AI & HEALTH · FOR PATIENTS
What AI Gets Wrong About Medications
Medication questions are where confident AI answers can do the most harm. A chatbot can explain how a medicine class works quite well — and in the next sentence invent a dose, overlook a dangerous interaction with your other medicines, or describe a drug that does not exist. This page covers the specific ways AI goes wrong on medications, and the rules that keep those errors from reaching your body.
By Dr. Mohamed Ibrahim, MD, PhD — Assistant Professor of Medicine, Transplant Nephrology, University of Maryland School of Medicine
✓ Medically reviewed by Mohamed Magdy Ibrahim, MD, PhD — Assistant Professor of Medicine, Transplant Nephrology, University of Maryland School of Medicine · Board certified in Internal Medicine (ABIM) · Last reviewed: November 2026
AI & Health is the MedicalVideos.com section on artificial intelligence in health care. We recommend no commercial product; where we test AI systems, we publish our method and our full results.
The guide
The picture today
Safe prescribing is personal. The right medicine and dose depend on your kidney and liver function, your other conditions, your age and weight, your allergies, and everything else you take — including over-the-counter products and supplements. A chatbot knows only what you type, and most people do not type a complete, accurate medicines list. Faced with gaps, the model does not stop; it fills them with plausible text. That is how invented doses, wrong frequencies, and missed interactions appear — written fluently, formatted neatly, and wrong.
Strengths and limits
The recurring failure modes are worth naming. Hallucinated details: doses, durations, or product names that sound right but are not. Missed interactions: the tool discusses one drug at a time and overlooks the combination that actually matters. Outdated or region-mixed guidance: approvals, brand names, and standard practice differ between countries, and answers blend them. False reassurance about supplements and herbal products, which interact with prescription medicines more often than people expect. And misplaced confidence about high-risk situations — pregnancy, breastfeeding, children, kidney or liver disease — where the margin for error is smallest.
In practice
Use AI for understanding, not instructions. It is reasonable to ask what a medicine is for, how its class works, what side effects are known in general, and what questions to ask your pharmacist. It is never reasonable to ask how much you should take or whether you can stop — and if it volunteers those answers anyway, treat them as unverified. Check anything important against the official patient information leaflet or a trusted drug reference, and put the question to your pharmacist, who can see your actual dispensing record.
Red flags
Some medication situations need a professional the same day, not a chatbot: a possible allergic reaction, signs of bleeding on a blood thinner, a missed or doubled dose of a critical medicine such as insulin or a transplant drug, or a new symptom that starts soon after a new medicine. Suspected overdose is an emergency — call your local emergency number or poison control service immediately. When a chatbot's advice conflicts with your prescription label, follow the label and contact your care team.
Common questions
The chatbot gave me a dose that matches my prescription. Is it fine then?
Agreement with your label is reassuring, but the label is the authority, not the chatbot. The risk is the day it does not match and sounds equally certain. Doses come from your prescription, your pharmacist, and official references — never from a generated answer.
Can AI check interactions between my medicines?
It can discuss interactions in general, but it cannot check yours: it does not have your verified medicines list, doses, or history, and it can miss combinations or invent ones that do not exist. Pharmacists run interaction checks against your real record — that is the check to rely on.
Is it safe to ask AI about side effects?
For general information, yes — what side effects a medicine is known to cause, and which ones warrant prompt review. But AI cannot judge whether your symptom is caused by the medicine, something else, or both. New or worrying symptoms after starting a medicine deserve a call to your care team, not just a chat.
What about vitamins and herbal supplements?
Treat them as medicines for safety purposes, because your body does. Supplements can change how prescription drugs work, and chatbots often understate this. Tell your clinician and pharmacist everything you take, and be wary of any source that calls a product 'natural and therefore safe'.
More in For Patients
Related
Sources
- Evaluating Health Information (MedlinePlus, US National Library of Medicine)
- Ethics and Governance of Artificial Intelligence for Health (World Health Organization)
- Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical Devices (US Food and Drug Administration (FDA))
This page is educational content about artificial intelligence and health. It is not medical advice and does not replace your own clinician's judgment. Never use an AI tool in an emergency: if you think you may be having a medical emergency, call your local emergency number (911 in the United States) or go to the nearest emergency department. Nothing on this page is a dosing reference; our Drug Guide deliberately publishes no doses either, for the same reason.