A carregar...
Publicação
Deepfakes e prova digital na violência doméstica: os riscos da inteligência artificial
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 721.03 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Contextualização: Os avanços recentes da inteligência artificial generativa têm introduzido novos desafios ao sistema de justiça penal, em particular no domínio da prova digital. Entre as aplicações mais problemáticas destacam-se os deepfakes, suscetíveis de induzir em erro quanto à sua autenticidade, especialmente em contextos de maior fragilidade probatória, como os crimes de violência doméstica. Objetivos: O presente estudo tem como objetivo geral compreender as perceções, experiências e práticas dos/as magistrados/as relativamente à admissibilidade e valoração da prova digital, com especial enfoque nos meios de prova audiovisual e nos riscos associados à sua eventual manipulação por tecnologias de inteligência artificial generativa. Métodos: Foi adotada uma abordagem qualitativa, recorrendo-se à realização de entrevistas semiestruturadas a oito magistrados/as, com experiência profissional em processos de violência doméstica. Resultados: Observaram-se limitações na deteção de elementos audiovisuais gerados ou manipulados por inteligência artificial. A autenticidade deste tipo de elementos (deepfakes) granjearam mais credibilidade que elementos autênticos. De todo o modo, embora a prova audiovisual seja considerada relevante, o depoimento da vítima continua sendo considerado a “prova rainha”.
Contextualization: Recent advances in generative artificial intelligence have introduced new challenges to the criminal justice system, particularly within the realm of digital evidence. Notable among the most problematic applications are deepfakes, which are liable to mislead regarding their authenticity, especially in contexts of significant evidentiary fragility, such as domestic violence crimes. Objectives: The general objective of this study is to understand the perceptions, experiences, and practices of magistrates regarding the admissibility and assessment of digital evidence, with a special focus on audiovisual evidence and the risks associated with its potential manipulation by generative artificial intelligence technologies. Methods: A qualitative approach was adopted, utilizing semi-structured interviews with eight magistrates possessing professional experience in domestic violence proceedings. Results: Limitations were observed regarding the detection of audiovisual elements generated or manipulated by artificial intelligence. These types of elements (deepfakes) garnered more credibility than authentic elements. Nevertheless, although audiovisual evidence is considered relevant, the victim's testimony continues to be regarded as the "queen of proofs".
Contextualization: Recent advances in generative artificial intelligence have introduced new challenges to the criminal justice system, particularly within the realm of digital evidence. Notable among the most problematic applications are deepfakes, which are liable to mislead regarding their authenticity, especially in contexts of significant evidentiary fragility, such as domestic violence crimes. Objectives: The general objective of this study is to understand the perceptions, experiences, and practices of magistrates regarding the admissibility and assessment of digital evidence, with a special focus on audiovisual evidence and the risks associated with its potential manipulation by generative artificial intelligence technologies. Methods: A qualitative approach was adopted, utilizing semi-structured interviews with eight magistrates possessing professional experience in domestic violence proceedings. Results: Limitations were observed regarding the detection of audiovisual elements generated or manipulated by artificial intelligence. These types of elements (deepfakes) garnered more credibility than authentic elements. Nevertheless, although audiovisual evidence is considered relevant, the victim's testimony continues to be regarded as the "queen of proofs".
Descrição
Palavras-chave
Prova audiovisual Valoração Deepfakes Autenticidade Inteligência artificial Audiovisual evidence Evidentiary assessment Authenticity Artificial intelligence
Contexto Educativo
Citação
Editora
UNIVALI