It took 5 days, numerous reports, and a formal complaint before YouTube shut down the channel behind a series of deepfakes — AI-generated videos that misused the name, image, and voice of Serge Hercberg, French physician and nutrition specialist best known for codeveloping Nutri-Score, and made him appear to say fabricated things.

Hercberg decided to maintain his complaint against the channel to raise awareness of the growing risks of health misinformation. Backed by several colleagues, he is calling for more effective safeguards to protect scientific speech.

What Happened

A few days earlier, Hercberg had learned of several deepfakes posted on a YouTube channel, in which an AI-generated figure bearing his likeness used his name and voice. Seated at a desk and wearing a white coat, the fake Hercberg dispensed bogus nutrition advice on slowing dementia, recommended avoiding certain vitamins to protect the liver and heart, and praised three plant proteins for repairing worn knee cartilage.

Created on December 31, 2025, the Santé Sage channel posted 32 videos with false content using Hercberg’s image without ever stating that they had been created with AI.

“I don’t know what motivated the people behind these videos — whether there was a commercial purpose or an intent to cause harm,” Hercberg told Medscape’s French edition.

A Broken Reporting Process

Hercberg filed several reports with YouTube and the French government’s PHAROS platform, an official portal for reporting illegal or harmful online content, and lodged a complaint against Santé Sage.

After 5 days of unsuccessful efforts and several refusals from the platform, YouTube finally removed the Santé Sage channel on February 10. Hercberg nonetheless intends to maintain his complaint in order to secure a conviction against the people behind the deepfakes, which harmed his reputation.

Mathieu Molimard, author of a recent report on health misinformation, has spoken out to denounce the shortcomings of platform moderation. “This case shows that the reporting process does not work,” Molimard, who heads the Department of Medical Pharmacology at Bordeaux University Hospital in southwestern France, told Medscape’s French edition. “It took escalation at the highest level to get this channel shut down. Moderation is not a priority for platforms, which either do not do this work or do it badly because it runs counter to their business model, which depends on video views. Today, disinformers and charlatans can act with complete impunity and disparage scientists because they run no risk. That has to change.”

In a post published on LinkedIn just after the video was taken down, Molimard said it was imperative to put in place more effective safeguards, relying on genuine trusted flaggers: public agencies, medical societies, universities, and hospitals. Without strong action to prevent such incidents from happening again and without simplifying and humanizing appeals procedures, these platforms should be considered responsible — or even complicit — in this conduct.

A Global Phenomenon

Why did YouTube take so long to respond and shut down the channel despite the many reports? Contacted by Medscape’s French edition, YouTube Health executive Asmaa Zerkdi did not answer questions. “I had contact with the platform on the day the channel was shut down,” Hercberg said. “They told me they would investigate why the reporting process had failed.”

Deepfakes — a process that involves altering or creating videos with AI from real images — have been spreading for several years. Several physicians have already been targeted, including well-known French TV doctor Michel Cymes. After appearing in earlier videos promoting a weight-loss treatment, a fake Cymes was promoting a miracle syrup for diabetes in the fall of 2024.

Last July, Marine Lorphelin, general practitioner and former Miss France, also expressed frustration over AI-generated videos of her being used to sell dietary supplements. “I would never promote miracle foods or products whose effects have not been scientifically proven,” she said on Instagram.

The phenomenon is global. In Australia, science broadcaster Karl Kruszelnicki also saw his image hijacked in a deepfake used to sell miracle pills to Australians. In England, Michael Mosley, who died in June 2024, was also the subject of AI-generated fake videos in which he appeared to promote products said to treat hypertension and diabetes.

Threat to Patient-Physician Trust

Deepfakes are also on the radar of the French Medical Council, the profession’s national self-regulatory body. “We are working on the issue of deepfakes as part of a working group because we are confronting surprising developments in AI that could create real difficulties in the short term,” Jean Canarelli, the council’s general delegate for health data, digital technology, and innovation, told Medscape’s French edition.

He urged physicians targeted by these practices to report them to their local medical council in order to protect themselves. “Whatever form identity theft takes, we condemn it firmly because it harms physicians and the trust patients place in them. And a patient who does not trust their doctor is treated poorly,” Canarelli added, calling for the legal framework to be updated to punish disinformers.

Up to 2 Years in Prison and a 45,000-Euro Fine

The French Penal Code does provide for sanctions against videos or uses of images that amount to an attack on a person’s identity. “ Article 226-8 of the Penal Code punishes with 1 year in prison and a 15,000-euro fine the publication of an edited image or recording made using someone’s words or image without that person’s consent and without clearly stating that it is an edited production,” explained Renaud Budet, lawyer and head of the Paris Bar Association’s health commission.

“An edited production made using the words or image of a consenting person, or made explicitly as such, even with AI, is not in itself unlawful,” explained Arnaud Dimeglio, appeals court lawyer specializing in digital, communications, and intellectual property law, in an article published on the specialized legal blog Village de la Justice, a French legal news and commentary site. “What is illegal is, first, creating an edited production without the person’s consent and, second, concealing the fact that it is edited.”

Speaking to Medscape’s French edition, Dimeglio added that the SREN law — a recent French law intended to secure and regulate the digital space — increases the penalty to 2 years in prison and a 45,000-euro fine when the offense is committed using an online public communication service.

Removal Doesn’t End Legal Action

Although victims of deepfakes initially focus their efforts on getting the videos removed, they may then seek damages. “Removing the videos does not extinguish legal proceedings insofar as they created harm for the person concerned,” Dimeglio said. The network that distributed the deepfake video may also be prosecuted if it delayed removing it. One of the main difficulties is identifying the people behind the offense.

The digital law specialist regularly files petitions with the court president to identify the authors of defamatory reviews hiding behind anonymous accounts. “It is very important to file a prompt complaint against unknown persons in defamation cases because the statute of limitations is 3 months,” Dimeglio concluded.

This story was translated from Medscape’s French edition.