Half of Koreans who changed their medication on AI advice reported side effects
South Korea's consumer agency found that about half of the people who stopped or altered a prescription on chatbot advice got worse or reacted badly.
The short version
- The Korea Consumer Agency surveyed adults who had bought medicines or supplements that an AI recommended.
- Some stopped or changed a prescription without asking a professional; about half of them had side effects.
- Only a minority checked with a doctor or pharmacist before buying; some delayed or skipped a hospital visit.
- Two of the four chatbots tested gave drug information without asking about age, conditions or current medicines.
- Three chatbots explained how to buy medicines that are banned from online sale.
Why it matters
Health questions are among the most natural things to ask a chatbot, and the answers sound confident and personal. This survey suggests that confidence can lead people to act against their prescriptions. Treat AI as a way to prepare questions for your doctor, not to overrule one.
What the agency found
As Yahoo! News reported, the Korea Consumer Agency studied how people use generative AI (software that writes replies based on patterns it learned from text) for medicine and health supplements. It surveyed adults who had bought products an AI had suggested.
A notable group had stopped or changed a prescribed drug on the strength of AI information, without asking a professional first. About half of that group said they then had side effects, such as a worsening illness or an unusual reaction.
Skipping the doctor
The habit went beyond prescriptions. Only a minority of respondents spoke to a doctor or similar professional after reading AI answers and before buying a product. Some also put off a hospital visit or gave up on going at all.
The survey relies on what people reported about themselves. It shows a pattern worth taking seriously, but it does not prove that the AI caused each case of harm.
How the chatbots measured up
The agency also examined answers from four services: ChatGPT, Gemini, Claude and Perplexity. Two of them handed out drug information after a bare symptom description, without asking about the user's age, existing conditions or current medicines. Those are exactly the details a pharmacist would want first.
All four included a note advising users to consult a professional. But the note was in small type, or sat at the bottom of the screen in the same style as the main text, so it was easy to miss.
A separate risk: restricted drugs
Three of the four services also gave web addresses and instructions for buying medicines that cannot legally be sold online, through purchasing-agent sites. An agency official said this could lead consumers to misuse or abuse illegal drugs.
What to watch
- Will the chatbot companies make their health warnings more prominent or ask users about their health first?
- Will regulators in Korea or elsewhere act on the finding that chatbots explained how to buy restricted drugs?
Try it
Before you stop, skip or change any medicine because of something a chatbot said, call your doctor or pharmacist. Use the chatbot to write down your questions for that call.
Sources and links
- ReportingYahoo! News: the article we sourced
First reported by Yahoo! News. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.