- Teenagers across Brazil—including in Rio de Janeiro, Bahia, and Mato Grosso—have been reported for creating and sharing AI-generated non-consensual sexual imagery of classmates.
- Independent research center Internetlab has released a technical note categorizing such deepfakes as an "excessive risk" and calling for urgent regulatory intervention.
- Data from Security Hero indicates that 98% of deepfake videos online are sexually explicit, with women comprising 99% of the targets.
The Intersection of Gender Violence and AI
- Internetlab research director Clarice Tavares argues that online and offline violence are manifestations of the same systemic misogyny.
- AI models often rely on biased training data, which inherently mirrors and amplifies gender-based prejudices and violence.
- Embedded AI tools on social media platforms, such as X's Grok, are criticized for lowering the barrier to entry for the creation and widespread dissemination of harmful content.
Brazil's Legal Landscape
- The Brazilian Supreme Federal Court (STF) recently ruled Article 19 of the 'Marco Civil da Internet'—the country's internet civil rights framework—partially unconstitutional, citing the need for greater platform accountability.
- Despite ongoing debates over legislative criminalization of misogynistic behavior, experts warn that relying solely on criminal law limits the efficacy of victim reparation and prevention strategies.
Policy Recommendations
- Implementation of "safety by design" principles is recommended to prevent the generation of harmful content at the software conception phase.
- Promotion of digital literacy programs to foster critical analysis of algorithms and AI ecosystems.
- As Brazil approaches election cycles, concerns are growing regarding the use of AI tools like Gemini, ChatGPT, or Claude to propagate gender-based political violence.