Analysis of Tort Liability of Generative Artificial Intelligence Service Providers

数字经济飞速发展中，以ChatGPT为代表的生成式人工智能既为人类带来益处，也不可避免的为法律和伦理带来挑战。现阶段的技术水平下，生成式人工智能不能确立其法律主体地位，而应由其服务提供者承担法律责任。对服务提供者认定侵权责任，首先要确定其属于新型网络服务提供者，然后根据生成式人工智能的训练、生成和移除阶段分别适用过错推定原则、无过错责任原则、过错责任原则的归责原则。以相当因果关系认定服务提供者侵权的因果关系。在保护被侵权人利益的同时，为了鼓励创新，可对服务提供者提供保护措施，以受害人故意和不可抗力的免责事由和服务提供者尽到注意义务的减责事由进行保护，有利于科技创新的发展。In the rapid development of the digital economy, generative artificial intelligence represented by ChatGPT not only brings benefits to humanity but also inevitably poses challenges to law and ethics. At the current technological level, generative artificial intelligence cannot establish its legal subject status, and instead, its service providers should bear legal responsibility. To determine the tort liability of a service provider, it is first necessary to determine that it belongs to a new type of network service provider. Then, the principle of presumption of fault, the principle of no-fault liability, and the principle of fault liability should be applied respectively according to the training, generation and removal stages of generative artificial intelligence. Determine the causal relationship of the infringement by the service provider based on a considerable causal relationship. While protecting the interests of the infringed party, in order to encourage innovation, protective measures can be provided for service providers. Protection can be carried out based on the exemption grounds of the victim’s intentional act and inability to resist and the mitigation grounds of the service provider’s fulfillment of the duty of care, which is conducive to the development of scientific and technological innovation.