Os desafios no cumprimento das obrigações na era da inteligência artificial (IA): a limitação convencional da responsabilidade civil na utilização destes novos mecanismos

The article addresses the transformative role of Artificial Intelligence (AI) across various sectors, emphasizing its growing relevance and the legal challenges arising from its application. Tools such as chatbots, advanced systems like ChatGPT, and automated translation software exemplify this "golden era" of AI, offering benefits such as increased efficiency, error reduction, and cost savings. However, these advancements also entail risks, including violations of intellectual property rights, privacy concerns, and operational issues, raising questions within the domain of civil liability.The article focuses on the difficulty of adapting Portuguese and European legal frameworks to the reality of AI, particularly regarding the limitation of liability in contracts between users and providers or those executed through AI systems. Fundamental questions arise: Can civil law consider AI systems capable of discerning between lawful and unlawful actions? Is it possible to assign legal liability to such systems? If not, upon whom would the responsibility fall? Are existing legal norms adequate to address the specificities of this technology?The research seeks legal solutions that overcome the insufficiencies of current regimes, balancing the rights of contracting parties with the risks posed by AI. Through a documentary methodology, including analyses of legal doctrine, case law, and, where applicable, comparative data, the study concludes that Portuguese law faces significant limitations. Article 809 of the Civil Code, which governs clauses excluding liability, invalidates those involving intent or gross negligence. According to Pinto Monteiro, limitation clauses aim to restrict, in advance, liabilities that, in the absence of contractual provisions, would fall on the obligor. However, they face stringent restrictions imposed by case law, including rulings from the Supreme Court of Justice.Furthermore, European regulations, such as those on defective products, are inadequate to address the complexities of AI. The Artificial Intelligence Act (AIA), a proposal by the European Union, constitutes an embryonic framework but remains insufficient to regulate the legal challenges posed by AI systems. The article concludes that updated legal solutions are essential to address the specificities and innovations of this disruptive technology, ensuring a fair balance between contractual autonomy and legal certainty.