Abstract：The development and application of generative artificial intelligence, which presents powerful human-like technology, has led to artificial intelligence risks, the biggest of which is the risk of infringement. The tort of generative artificial intelligence has a particularity, the tort subject has diversity, the imputation principle has diversity, and the legal liability has dispersion. Generative AI does not qualify as a legal subject, and its service provider should bear the liability for infringement. The tort liability of generative AI service providers should be divided into training stage, generation stage and removal stage, and different liability principles should be applied according to the degree of risk. The identification and commitment of tort liability of generative artificial intelligence should be changed from the approach of liability form to the analysis paradigm of infringement scenario. Product-component generative AI applies product liability to its producers, service providers are treated according to fault liability, and punitive damages should be introduced to high-risk generative AI products. The generative AI of auxiliary decision-making should compare the fault allocation of tort liability between the user and the service provider, and assign the exemption of liability. The tort of generative artificial intelligence and large model involves the liability of service provider, platform publisher, etc., and the level and size of liability should be distinguished to achieve a balance between risk and liability.
2024, Vol. 41 
粤公网安备44030502007936号