Meeting the challenge of health misinformation together

False and misleading information is not new, but in today’s information environment it can spread faster and further than ever before. People are constantly exposed to seemingly scientific claims which are unintentionally misleading or deliberately false through social media, from influencers, via search engines and, increasingly, by AI-generated tools.

The consequences can be severe 

In the Democratic Republic of Congo, false claims about Ebola and the motivations of health professionals have helped fuel attacks on treatment facilities, assaults on health workers and repeated attempts to interfere with safe burial procedures. 

But this is not only a problem overseas or in moments of crisis. In the UK, for example, clinicians are increasingly having to respond to misleading online claims about food supplements and dietary approaches linked to cancer prevention and treatment¹, with patients often not made aware of possible risks online, including side effects and interactions with other medicines.

Similar dynamics can be seen in women’s health, where false or exaggerated claims about hormonal contraception can shape important personal choices, particularly where they connect with people’s real experiences of side effects and feeling that medical professionals are not listening to them². 

These examples are different in scale and context, but they point to the same underlying issue: people naturally look for answers when worried about their health, but the quality of answers available online, whether from influencers, search engines or from generative AI, can vary considerably.

Not all misleading information is ill-intentioned. Some misleading claims spread because people are anxious, trying to help, or drawing on personal experiences. But whatever the motive, the effect can be the same: confusing or inaccurate information reaches people at the moment they are trying to make important decisions.  

Consumer-facing AI tools could potentially help people to understand straightforward health concerns and identify self-care measures³. However, their performance varies considerably, particularly on diagnosis^4,5. They can also provide inaccurate or poorly referenced answers with unwarranted confidence, particularly when asked challenging questions in areas already prone to misinformation⁶.  

AI-generated information, therefore, cannot replace professional judgement. The performance of public-facing health tools needs to be rigorously evaluated, and they should state their level of certainty, ideally in quantitative terms, which will allow them to be assessed.

While misinformation matters, it also cannot become a lazy diagnosis of challenges in relation to healthcare delivery. If we assume misinformation explains every mismatched public response to evidence or health programmes, we risk missing the wider reasons people make the decisions they do, from lived experience to practical barriers affecting access. We need to understand misinformation clearly and build responses that are proportionate and based on sound evidence.

This is one of the reasons why the Government Chief Scientific Adviser has made trustworthy information her priority this year. It is essential that the study of misinformation is evidence-based so that government and the wider science system can adapt to the changing information environment without overstating the problem, misdiagnosing its causes, or overlooking interventions likely to help.  

A recently completed evidence review by Behavioural Research UK underlines that false and misleading information is a systemic risk, but its impacts can be difficult to measure and no single intervention is enough on its own – mitigation requires layered, adaptive strategies, combining individual and system-level actions.

The medical sciences community has a significant role to play

Researchers, clinicians, funders, journals and learned societies are all actors in the information environment. They shape what evidence is produced, how it is explained, and how uncertainty is handled. Such influence brings responsibility.

First, trusted information needs to be available early, and in forms people can use. Too often, good evidence exists but does not answer the practical questions people are asking: How serious is the risk for someone like me? How strong is the evidence? What are the benefits and harms? What choices are available to me?  

We should also distinguish between people with genuine concerns and those with more malign intent. Most people are not trying to spread falsehoods. They are trying to make sense of complex information, often in contexts where decisions feel personal or urgent. They deserve clear, respectful answers.  

At the same time, those deliberately peddling mis- and disinformation want us to believe they represent the majority. They do not. We should avoid giving them more credibility or reach.

How we respond matters. We should avoid giving false claims unnecessary prominence, while still being clear about what the evidence shows. Simply labelling something as false is rarely enough: people need an explanation that addresses the often valid concerns behind the claim and provides accurate, accessible information in its place. The right response will depend on the audience, the reach and context of the claim and the risk of harm.