AI videos of children appear in Meta ads, including Danish victim

DR reports that AI-generated videos depicting children in sexual situations are being used to advertise nudify apps on Meta’s platforms. One of the children identified by the broadcaster is Danish.

DR’s verification team followed social media advertisements featuring pornographic or sexualised content, often created from real people’s images without their consent. Meta, the company behind Facebook and Instagram, charges advertisers to distribute the material on its platforms.

The investigation found that images of children were also used to create AI-generated abuse material. Some of it was then used to promote nudify apps, which allow users to turn ordinary photographs into pornographic videos.

Miriam Michaelsen, chair of Denmark’s Media Council for Children and Young People, called the presence of advertisements containing sexualised images of children “insane”. She said the material was clearly illegal and offensive both to those depicted and to people who encounter it without seeking it.

Meta did not respond to several interview requests. In an email, the company said it does not allow the promotion of nudify apps and had removed many of the advertisements presented by DR.

“We do not tolerate nudify apps or any form of sexual exploitation, whether real or AI-generated, and we are taking firm action to keep our platforms safe,” Meta said.

With help from Red Barnet’s AnmeldDet reporting service, DR reviewed nine advertisements suspected of containing sexualised material involving minors or AI-generated abuse material.

According to Red Barnet, four of the advertisements initially show images of children before AI changes them into videos depicting explicit sexual activity. AnmeldDet reported all four videos to the police as abuse material, potentially violating section 235 of Denmark’s Penal Code.

Section 235 prohibits the distribution of photographs, films or other visual representations containing sexual content involving people under 18. The offence carries a prison sentence of up to two years, or up to six years in particularly aggravating circumstances. In Denmark, the law applies whether the children are real or depicted in AI-generated images.

In one of the four cases, the victim was a Danish child whom DR was able to identify. DR contacted the child’s parents, who did not wish to speak publicly or comment on the case. Another identified child was from Australia.

“If the people shown are under 18 and involved in any form of sexual activity, we refer it to the police so they can investigate,” said Pernille Spitz, Red Barnet’s head of protection against abuse.

Spitz said the material she reviewed appeared highly realistic and involved what appeared to be very young girls in heavily sexualised situations.

Other advertisements reviewed by Red Barnet did not immediately appear to violate the law on child abuse material, but still had a sexualised expression. They used features commonly associated with children, including pigtails, clothing featuring the cartoon character Hello Kitty and an innocent facial expression.

The advertisements found by DR also attracted users by promising that there were no restrictions. Michaelsen said deepfake technology was becoming increasingly advanced, making the violations more extensive and severe. She called for stronger accountability across the technology sector.