Help that is always available
The paper recognises why chatbots have become popular sources of health information. They are available at any hour, free or cheap, easy to use and able to explain medical concepts in plain language, which can help people prepare for appointments or understand diagnoses.
But the authors note that generative AI can produce inaccurate information with confidence and can also be used to generate personalised misinformation that spreads quickly on social media, reaching people who were never looking for health advice in the first place.
Harm that leaves no trace
The central warning concerns visibility. Unlike a doctor's consultation or a medicine's side effect, a harmful chatbot answer leaves no trace in health records or official reporting systems.

















