Are AI chatbots safe for medical advice?

Evidence, risks and practical precautions

Multiple studies and reporting indicate that popular AI chatbots are not yet reliable substitutes for professional medical care. Research published in prominent journals and news coverage find that large language models can produce incorrect, misleading, or potentially harmful recommendations when asked for medical guidance. In several accounts, people said conversational agents gave dangerous advice; other studies documented models echoing medical misinformation seen on social media.

The reasons are straightforward: these systems generate plausible‑sounding text based on patterns in their training data rather than clinical reasoning, and they lack consistent access to verified, up‑to‑date medical records or real‑time clinical context. In low‑resource settings, people are turning to chatbots because they cannot easily access or afford therapists and clinicians — a trend that increases exposure to unreliable advice.

What patients and clinicians should do - Treat chatbot responses as informational only; verify with a licensed clinician before acting on medical advice.
- Avoid using chatbots for emergency or urgent decisions; call emergency services when required.
- Ask for sources and second opinions when a chatbot offers factual claims.

Open questions Researchers are still testing how well specific models perform across different medical scenarios and what standards should govern their use. Policymakers and medical groups are calling for tighter oversight and clearer warnings about limits. Until more robust validation and regulation are in place, the safest course is to use AI tools as a supplement — not a replacement — for professional medical assessment.