On the screen of a phone, an 82-year-old woman watches a man in a white coat speak with confidence about remedies and cures. The voice is steady and the advice sounds reasonable. "I didn't realize it was artificial intelligence. I thought it was a real doctor, because what he said seemed plausible," she told BBC News Brasil. The doctor on screen, however, does not exist: the face, the voice and the coat were manufactured by artificial intelligence.
That scene is what the Brazilian federal government has now moved against. The Advocacia-Geral da Uniao (AGU), the office that represents the federal state in court, notified YouTube on Friday (4 September) to remove or flag videos that use artificial intelligence to create fake doctors and other health professionals spreading misleading claims, such as promises of cures and scientifically unproven treatments. The platform has 72 hours to make the contents identified by the Ministry of Health unavailable or, depending on the case, to adopt labels and warnings making clear that the characters are artificial. The notification draws on an analysis by the ministry's Saude com Ciencia (Health with Science) program, which identified 97 profiles publishing this type of content between January 1 and July 10. The material was also sent to the Federal Police and to the Conselho Federal de Medicina (CFM), Brazil's national medical council.
According to the ministry, the videos present human-looking avatars who introduce themselves as doctors or specialists, carry fictitious professional credentials and rely on alarmist language to build credibility. Some of the profiles target older Brazilians specifically and use the characters' supposed authority to sell products, e-books and paid services. The ministry warns that such content can lead to self-medication, use of therapies with no proven efficacy and delays in seeking professional care, including patients interrupting or replacing prescribed treatments. YouTube was also urged to review the other channels and videos named in the report under its own rules on health misinformation and synthetic content.
An industry at scale
The notification comes two months after BBC News Brasil exposed an industry of channels built on these characters. The July investigation mapped 29 Portuguese-language channels devoted to AI fake doctors with more than 70 million combined views, based on data from the nonprofit CTRL+Z, and counted roughly ten new videos per day. In about 27,000 comments, the BBC found accounts from older viewers who said they had followed the advice, including people who replaced prescribed medication with foods, plants and home remedies. Contacted at the time, Google removed some of the largest channels in the survey, with about 41 million combined views, and said all AI-generated content must follow guidelines that ban medical misinformation capable of causing serious real-world harm. Other videos remained online, and the BBC says it is still waiting for a response from the company on the new notification.
When the face belongs to a real doctor
A second BBC report showed that at least five channels used, without authorization, the face and identity of otolaryngologist Helio Brasileiro to recommend supposed natural treatments and make alarming health claims. He discovered the clones when he saw his own image in the first BBC report, complained to YouTube and filed a police report with the civil police of Sao Paulo state. The case is being investigated by the 3rd police district of Sorocaba, which is weighing possible offenses including ideological falsehood, false identity, false alarm and attempted defamation. In her order, delegate Alessandra Silveira wrote that spreading medical disinformation through biometric simulation demanded immediate police action and could create a risk of poisoning or self-medication.
The notification to YouTube is part of a task force that includes the Ministry of Health, the AGU and the presidential communications office. The ministry recommends that Brazilians check health guidance against official sources and report content that could put other people at risk. For the 82-year-old woman who believed in the doctor on her screen, what remains is a warning that should have arrived sooner: that doctor never existed.