Misleading AI-generated doctors pose ‘huge danger to public safety’

Misleading AI-generated doctors pose a “huge danger to public safety,” according to experts following new research revealing how artificial intelligence tools are being used to create fake medical content that spreads unverified health claims. A study by Hallam, a digital marketing agency, found that AI-generated physician profiles on TikTok are reaching millions of viewers, often promoting debunked medical theories and dubious treatments. The research highlights the alarming scale of the issue, showing that AI-generated content appears in 40% of top health-related TikTok videos. In some cases, up to 84% of videos for specific search terms, such as “health tips,” were classified as AI-generated or AI-assisted. These videos are typically created by deepfake influencer accounts designed to mimic real doctors, raising concerns over the credibility of the information being shared. Dr Emma Runswick, deputy chair of the British Medical Association council, warned that these accounts “peddle medical myths and promote so-called miracle cures.” She emphasized the need for platforms like TikTok to take stronger action against the proliferation of fake medical advice, stressing that the consequences could be severe for the public's health and even life-threatening. The AI-generated content includes claims that have been thoroughly debunked by reputable organizations. For instance, it suggests that microwaving food in plastic, using deodorants, and sleeping next to a phone can lead to cancer, claims that Cancer Research UK has refuted. Additionally, the videos recommend unconventional remedies, such as mixing Coca-Cola with onions for pain relief and promoting a non-existent product named “Hyalethinap Plus Pro Max” as an anti-ageing solution. One AI-generated doctor claimed to have prescribed “Hyalethinap Plus Pro Max” to patients for years, asserting that it could address issues like hair loss, weak skin, brittle nails, and joint pain by stimulating the production of “anti-ageing molecules naturally.” However, the Guardian was unable to locate any evidence of the product online, suggesting that these recommendations are entirely fabricated. NHS England’s national medical director, Prof Frankie Swords, expressed concern over the surge in online medical misinformation. She noted that many healthcare professionals are encountering patients who are anxious and confused due to false advice tailored for maximum engagement rather than accuracy. Similarly, Prof Victoria Tzortziou Brown, president of the Royal College of GPs, echoed these sentiments, stating that misleading AI content poses a genuine risk of harm. Health misinformation researcher Alex Ruani from UCL described the situation as a “frightening amount of exposure to fake medical content.” She pointed out that the rise of AI doctor videos represents an “industrialised exploitation of trust,” highlighting the potential for widespread misinformation. The study analyzed 1,198 videos, focusing on the top 50 for search terms like “health tips” and “doctor advice.” It also examined broader topics such as personal finance and education. According to a poll conducted last year by Savanta on behalf of Healthwatch England, approximately one in five individuals in England uses social media for health information. William Pett, interim director of policy and public affairs at Healthwatch England, warned that without the ability to discern credible information from commercial hype or misinformation, users face serious risks to their health. TikTok responded to the findings, stating that they do not reflect an accurate portrayal of the platform. The company reiterated its commitment to removing harmful health misinformation, including AI-generated content, and highlighted partnerships with the World Health Organization and the NHS to provide reliable health information within the app. It also mentioned ongoing investments in AI literacy resources and labeling technologies aimed at improving user understanding of content authenticity.