Responsabilidade civil pelo fato imprevisível do Sistema de Inteligência Artificial: análise à luz do Código Civil de 2002 e do Código de Defesa do Consumidor

• Ano de defesa: 2024
• Autor(a) principal: Sousa, Alan Araújo de
• Orientador(a): Queiroz, Odete Novais Carneiro
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Pontifícia Universidade Católica de São Paulo
• Programa de Pós-Graduação: Programa de Pós-Graduação em Direito
• Departamento: Faculdade de Direito
• País: Brasil
• Palavras-chave em Português: Responsabilidade civil; Inteligência artificial; Fato imprevisível; Código Civil; Código de Defesa do Consumidor
• Palavras-chave em Inglês: Civil liability; Artificial intelligence; Unpredictable behaviour; Civil Code; Consumer Protection Code
• Área do conhecimento CNPq: CNPQ::CIENCIAS SOCIAIS APLICADAS::DIREITO
• Link de acesso: https://repositorio.pucsp.br/jspui/handle/handle/42756

This dissertation aims to examine, in light of the 2002 Civil Code and the Consumer Protection Code, the civil liability of individuals who provide products and services incorporating artificial intelligence (AI) systems, particularly for damage caused by their unpredictable behavior or results, in order to determine whether current regulations adequately ensure compensation of the damage suffered by the victims. Initially, the dissertation explores the main aspects of AI, a prominent technology in the context of the fourth industrial revolution, and its implications for damage compensation. It outlines the main challenges in the scope of civil liability, especially in terms of proving negligence, the causal link and demonstrating the defect when required, particularly in cases involving damage caused by the unpredictable behavior of AI systems. To find potential solutions, the study first examines the civil liability framework established by the 2002 Civil Code, focusing on tort liability, its foundations (negligence theory and risk theory) and an analysis of the general liability clauses found in articles 927, sole paragraph, and 931 of the 2002 Civil Code. The research then turns to the civil liability system under the Consumer Protection Code, providing a brief overview of the legal protections available to consumers and focusing on the rules relating to supplier liability for consumer accidents. This creates a basis for examining one of the pivotal issues of the research, which is the legal classification of the unpredictable behavior of AI-operated products and services: are they an inherent risk, a defect, developmental risk, developmental gap or an internal fortuitous event? Subsequently, the dissertation delves deeper into the study of defects, developmental risks, and analyzes civil liability for damage caused by the unpredictable behavior of products and services that incorporate artificial intelligence. This analysis is initially conducted under the Consumer Protection Code, and then under the general strict civil liability clauses in the 2002 Civil Code. At this stage, a brief examination of the issue from the perspective of causal link is also carried out. Finally, since artificial intelligence and its implications have already garnered the attention of legislators, the study concludes with a presentation of the standards and bills that aim to regulate this technology in both Brazilian and European (European Union) contexts, primarily seeking to identify the possible consequences for civil liability arising from unforeseeable actions of AI systems integrated into products or services. The conclusion is that the Brazilian civil liability system generally provides satisfactory solutions to the issue of compensating damage caused by the unpredictable behavior of AI systems. In equitable relations, due to the general strict civil liability clauses enshrined in the 2002 Civil Code; and in consumer relations, due to the possibility of classifying these AI unpredictable behaviors, when they cause damage, as a defect or internal fortuitous event