Health misinformation in the age of AI

Why this matters

False health information has always spread: miracle cures, rumours about vaccines, fake warnings passed around family groups. AI changes the speed and the scale. Anyone can now produce a convincing article, a realistic image, a fake audio message in a familiar voice or a video of a doctor saying something they never said, in minutes and at almost no cost. When people act on false health information they may delay treatment, refuse effective care, take something harmful or lose trust in health workers.

Health departments, district health offices and hospitals are often the first place people turn to for the truth. This module is about responding well.

Know the terms

• Misinformation is false information shared by people who believe it. Most health rumours spread this way, often with good intentions.
• Disinformation is false information spread deliberately, to profit, to cause confusion or to damage trust.
• Deepfakes are AI-generated images, audio or video that make a real person appear to say or do something they did not. Voice cloning can imitate a known doctor or official in a voice note.

The distinction matters for tone. People sharing misinformation are usually worried, not malicious. Treat them with respect.

How AI changes health misinformation

• Volume. Many versions of the same false claim can appear at once, in several languages, making it look widely supported.
• Polish. AI text is fluent and confident, with invented references that look scientific. Poor spelling is no longer a reliable warning sign.
• Impersonation. Fake messages may carry the name, logo, letterhead or voice of a real hospital, department or doctor.
• Chatbot errors. People also ask general AI chatbots health questions and can receive hallucinated answers that are wrong but convincing, with no bad intent behind them at all.

Responding: a practical approach

1. Listen first. Encourage staff to report rumours they hear from patients and communities. Frontline workers often know about a rumour days before it reaches the media.

2. Check before you react. Confirm what the false claim actually says and where it is circulating. Do not repeat or forward the material while checking. If it impersonates your institution, keep a copy as evidence and report it through your department's channels.

3. Decide whether to respond. Not every rumour needs a public answer. Responding to something few people have seen can spread it further. Respond when a claim is spreading, could cause harm or uses your institution's name.

4. Lead with the truth. State the accurate information first, clearly and simply. Mention the false claim briefly, explain why it is wrong and then repeat the accurate message. Avoid making the false claim the headline.

5. Use plain language and the right languages. Write in Urdu and in local languages where needed, at a level everyone can follow. Short messages, simple visuals and voice notes can reach people that formal notices do not.

6. Use trusted messengers. Community health workers, local doctors, teachers and respected community figures are often more trusted than official statements. Equip them with clear, accurate messages.

7. Tell people how to check. Remind the public that your institution will only communicate through its known official channels. Tell them how to verify a message that claims to come from you.

Using AI to help, carefully

AI can help your team draft clear public messages, simplify technical language and prepare versions for different audiences. The usual rules apply:

• every medical fact must come from an authoritative source and be checked by a qualified clinician before release
• never let AI invent statistics, quotes or references
• check Urdu and regional language versions with a fluent speaker
• the final message is approved by a person with the authority to approve it

Protect your own channels

Make sure your institution's official accounts and contact details are clearly published and consistent. Agree who can post and who approves health messages. If your institution is impersonated, a quick, clear statement on your official channels helps people tell real from fake.

Try it

Find a widely shared health rumour from anywhere in the world that has been publicly debunked by a reputable source. Using only that public information, draft a short public message (under 100 words) that leads with the truth, mentions the false claim once and ends with how people can get reliable information. Then ask an AI assistant to simplify it and compare the two versions for accuracy.

Check yourself

1. What is the difference between misinformation and disinformation?
2. Why should a public response lead with the accurate information rather than the false claim?
3. Name two things that must happen before an AI-assisted health message is released.

Answers: (1) Misinformation is shared by people who believe it; disinformation is spread deliberately. (2) Repeating the false claim prominently can help it stick and spread; leading with the truth makes the accurate message the one people remember. (3) Any two of: medical facts checked against an authoritative source by a qualified clinician, no invented statistics or quotes, language versions checked by a fluent speaker, approval by an authorised person.