Detect Deepfakesby Resemble AI
Deepfake case study · Multi-modal

Are you concerned about online misinformation and deepfake?

AI-generated deepfake videos are impersonating medical experts to spread health misinformation and promote supplement sales on major social media platforms

Incident date
Dec 2025
Target
David Taylor-Robinson, Duncan Selbie, Tim Spector, Michael Mosley
Updated Aug 11, 2026 · 1 min read

What happened

Social media platforms, including TikTok, X, Facebook, and YouTube, have hosted hundreds of AI-generated deepfake videos featuring prominent doctors and health influencers. These videos manipulate real footage and audio of experts to falsely endorse products from a US-based supplement firm called Wellness Nest. The investigation, conducted by the factchecking organization Full Fact, revealed that the manipulated content is specifically targeting women experiencing menopause, suggesting unproven treatments for symptoms such as "thermometer leg."

The deepfakes utilize existing footage of professionals, such as Professor David Taylor-Robinson, whose likeness was harvested from a 2017 public conference and a parliamentary hearing. In these doctored clips, Taylor-Robinson’s image and voice were synthesized to promote specific supplements and, in some instances, to depict him making misogynistic comments. Other high-profile figures, including former Public Health England chief executive Duncan Selbie, Professor Tim Spector, and the late Dr. Michael Mosley, were also impersonated using similar techniques. While Wellness Nest stated that the videos were "100% unaffiliated" with their business and claimed they do not use AI-generated content, the prevalence of these videos has sparked significant concern regarding the ease with which AI can be used to exploit public trust in medical expertise. Victims of the impersonations reported that the process of having the deceptive content removed from social media platforms was slow and inconsistent, with some platforms initially failing to recognize the videos as deepfakes.

Sources