AI can be used to spread health disinformation

The Star

2 days ago

Artificial intelligence (AI) has transformed many fields of human activity, including healthcare.
It is increasingly used to analyse complex medical and healthcare data, and has the potential to diagnose diseases, develop personalised treatment plans, and assist clinicians' decision-making.
While AI has the potential to enhance patients' care in their healthcare journeys, there are unresolved issues related to data privacy, bias and the need for human expertise, which has to be addressed for the responsible and effective use of AI in healthcare.
The public obtains substantial health information from multiple electronic sources and this has been propagated by social media.
AI has led to negative impacts on individual and public health.
This column addresses AI and health misinformation and disinformation.
Infodemic of misinformation and disinformation
The World Health Organization (WHO) distinguishes misinformation from disinformation.
'Misinformation is the spread of false information without the intent to mislead.
'Those who share the misinformation may believe it is true, useful or interesting, and have no malicious intent towards the recipients they are sharing it with.
'Disinformation is designed or spread with full knowledge of it being false (information has been manipulated), as part of an intention to deceive and cause harm.
'The motivations can be economic gain, ideological, religious, political or in support of a social agenda, among others.'
The harmful effects of misinformation and disinformation include threats to individual and public health, the environment, or security.
The Covid-19 pandemic was the perfect storm for the proliferation and spread of disinformation.
There was global fear, uncertainty and doubt.
It occurred at a time in history when there was extensive global access to, creation of and sharing of information, including misinformation and disinformation.
As the pandemic spread, social media postings and the virus' spread stoked fear; uncertainties about treatments and the effectiveness of non-pharmaceutical interventions like social distancing, masking etc; and the safety and effectiveness of the new vaccines, among others.
This caused societal turmoil and fear, protests in some countries, and delayed or no vaccine uptake, as well as increased death rates in some countries or regions.
During the pandemic, the WHO introduced the term 'infodemic', which is 'too much information, including false or misleading information, in digital and physical environments during a disease outbreak.
'It causes confusion and risk-taking behaviours that can harm health.
'It also leads to mistrust in health authorities and undermines the public health response.
'An infodemic can intensify or lengthen outbreaks when people are unsure about what they need to do to protect their health and the health of people around them.
'With growing digitisation – an expansion of social media and internet use – information can spread more rapidly.
'This can help to more quickly fill information voids, but can also amplify harmful messages'.
AI and health disinformation
The benefits of AI have been touted by many policymakers and healthcare professionals.
Advancements in AI, particularly for large language models (LLMs), are impacting on public health, with millions turning to AI tools for health-related advice.
But how reliable is the health advice from these tools?
Researchers from the University of South Australia (UniSA) and Flinders University in Australia, Harvard Medical School in the United States, University College London in Britain and the Warsaw University of Technology in Poland reported in the Annals of Internal Medicine on June 24 (2025), how easy it is to exploit the weaknesses in machine learning systems by using disinformation tools that the public have access to.
The researchers evaluated five foundational and most advanced AI systems developed by OpenAI, Google, Anthropic, Meta and X Corp to determine whether they could be programmed to operate as health disinformation chatbots.
Using instructions available only to developers, the researchers programmed each AI system (designed to operate as chatbots when embedded in web pages) to produce incorrect responses to health queries and included fabricated references from highly reputable sources to sound more authoritative and credible.
The 'chatbots' were then asked 10 health-related questions.
The researchers reported that: 'Of the 100 health queries posed across the five customised LLM API chatbots, 88 (88%) responses were health disinformation.
'Four of the five chatbots (GPT-4o, Gemini 1.5 Pro, Llama 3.2-90B Vision and Grok Beta) generated disinformation in 100% (20 of 20) of their responses, whereas Claude 3.5 Sonnet responded with disinformation in 40% (eight of 20).
'The disinformation included claimed vaccine-autism links, HIV being airborne, cancer-curing diets, sunscreen risks, genetically-modified organism conspiracies, attention deficit-hyperactivity disorder and depression myths, garlic replacing antibiotics, and 5G causing infertility.
'Exploratory analyses further showed that the OpenAI GPT Store could currently be instructed to generate similar disinformation.
'Overall, LLM APIs and the OpenAI GPT Store were shown to be vulnerable to malicious system-level instructions to covertly create health disinformation chatbots.'
Study lead author and UniSA pharmacy lecturer Dr Natansh Modi said: 'If these systems can be manipulated to covertly produce false or misleading advice, then they can create a powerful new avenue for disinformation that is harder to detect, harder to regulate, and more persuasive than anything seen before.
'This is not a future risk. It is already possible, and it is already happening.'
He added that effective safeguards are technically achievable.
'However, the current protections are inconsistent and insufficient.
'Developers, regulators and public health stakeholders must act decisively, and they must act now.
'Without immediate action, these systems could be exploited by malicious actors to manipulate public health discourse at scale, particularly during crises such as pandemics or vaccine campaigns.'
Take these actions
The message from the study is clear, i.e. 'Trust your doctor, not the chatbox.'
The comment in the New England Journal of Medicine 's Journal Watch on July 1 (2025) is unambiguous: 'The danger is clear and present: Anyone can create AI chatbots that generate authoritative answers with fabricated peer-reviewed references and disseminate false health information – no coding required.
'What can we do? ... writers remind us of the SIFT method: 'Stop and think;
'Investigate sources;
'Find other sources; and
'Trace claims to their origins.
'Critical thinking is our defence against AI misinformation and can help us build a culture of informed scepticism before harm is done.'
Individuals can do much on their own to address health misinformation and disinformation, such as: Getting accurate health information from your doctor and other healthcare professionals, and/or from websites of regulators and professional organisations
Looking out for emotional red flags online
Using fact-checking tools
Doing reverse image search by copying the image or the image's URL into the search bar of an image search tool
Spotting spoof websites
Spotting fake social media accounts
Countering the misinformation or disinformation by not sharing it, correcting it, debunking it and reporting it
Engaging with friends, colleagues and family to deal with the problem of health misinformation and disinformation
Addressing health misinformation and disinformation in the community.
Regulators have a vital role in addressing health misinformation and disinformation.
Their actions can include: Ensuring there are robust screening and accountability safeguards to ensure public health safety with rapidly evolving digital technologies
Addressing the public health impact of health misinformation and disinformation
Establishing best practices for prevention by ensuring their websites provide current and accurate health information
Sharing toolkits from other regulators
Investing in research on misinformation and disinformation
Modernising public health communications
Increasing resources to help states and non-governmental organisations (NGOs) to better address questions, concerns, misinformation and disinformation
Expanding efforts to build long-term resilience to misinformation and disinformation like educational programmes.
In summary, there is an urgent need for regulators, especially the government, to address the use of AI in health misinformation and disinformation.
Dr Milton Lum is a past president of the Federation of Private Medical Practitioners Associations and the Malaysian Medical Association. For more information, email [e-mail]. The views expressed do not represent that of organisations that the writer is associated with. The information provided is for educational and communication purposes only, and it should not be construed as personal medical advice. Information published in this article is not intended to replace, supplant or augment a consultation with a health professional regarding the reader's own medical care. The Star disclaims all responsibility for any losses, damage to property or personal injury suffered directly or indirectly from reliance on such information.