SaferNet maps 173 sexual deepfake victims in Brazilian schools
SaferNet Brasil documented 173 women and girls who were victims of non-consensual sexual deepfakes created with generative AI in schools across ten Brazilian states. The mapping found all victims were female and identified 60 perpetrators operating through organized networks on Telegram and dark web forums.
What happened
A mapping by SaferNet Brasil identified 173 victims of sexual deepfakes in public and private schools across ten Brazilian states. All identified victims are women, including students and teachers, with São Paulo leading at 51 cases, followed by Mato Grosso (30), Pernambuco (30), and Rio de Janeiro (20).
Sexual deepfakes are images or videos of nudity created using generative artificial intelligence without the consent of the people depicted, manipulating their faces into false content. The study began in 2023 based on news monitoring and is funded by SafeOnline, managed by Unicef. The complete report is scheduled for release in March.
Reported impact
The deepfakes constitute a violation of privacy and human dignity. SaferNet identified 60 perpetrators of these crimes and noted a 28% increase in cybercrime reports in 2025.
The Central Nacional de Denúncias de Crimes Cibernéticos received 264 links related to this type of crime since 2023, of which 8% involved artificial content depicting sexual abuse and exploitation of children.
Index analysis
Criminal groups sharing this content operate in an organized manner using automated notification bots, Telegram channels, and dark web forums. These groups exploit governance failures in digital platforms and state oversight.
Governance controls
SaferNet Brasil calls for the banning of notification tools and the financial asphyxiation of these criminal networks to combat the spread. The organization operates the Central Nacional de Denúncias de Crimes Cibernéticos, which accepts anonymous reports about cybercrime including sexual abuse of children.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.