Translated with AI
Millions of children and adolescents worldwide are becoming victims of crimes related to the creation of sexualized deepfakes, according to a joint study by UNICEF, the international organization ECPAT, and Interpol, conducted in 11 countries, the UN press service reported.
According to the study, approximately one in 25 children falls victim to perpetrators creating sexualized deepfakes. At least 1.2 million children reported that their images had been manipulated and transformed into sexually explicit fakes. UNICEF emphasized that the scale of the problem is comparable to a situation where at least one child in every school classroom has been a victim of such a crime.
It is noted that deepfakes—photos, videos, or audio files created or processed using artificial intelligence technologies that look or sound like the real thing—are increasingly being used to create sexualized content featuring children.
The organization noted that children themselves are aware of the existing risks. In several countries participating in the study, approximately two-thirds of children said they were concerned about the possibility of artificial intelligence being used to create fake sexual images or videos.
UNICEF emphasizes that the creation of sexualized deepfakes amounts to child abuse. Experts believe such fake videos and photographs normalize the sexual exploitation of minors and fuel demand for violent content.
The foundation welcomed the efforts of AI developers to implement safe-by-default approaches and robust security mechanisms to prevent misuse of their systems. However, it emphasized that many AI models are being created without sufficient security measures, and risks increase when generative AI tools are directly integrated into social media, where processed images are shared at high speed.
UNICEF calls on all states to criminalize the creation, acquisition, storage, and distribution of AI-generated content containing sexualized images of children. Furthermore, the organization believes it is essential for AI developers to implement safety mechanisms at the design stage, and for digital companies to prevent the dissemination of such material, rather than simply removing it after it has been abused.
