Liability Principles for Copyright Infringement of Generative Artificial Intelligence
<p>Chen Yinfei</p> · The Frontiers of Society Science and Technology · 2025
Unlike the traditional network copyright infringement model, the copyright infringement model of Generative Artificial Intelligence is more complex. In this model, the infringing subject cannot be simply defined as direct infringement or indirect infringement. For network users, if they consciously input instructions to guide the generation of infringing content, the principle of fault liability should be adopted. For the providers of Generative Artificial Intelligence services, their copyright infringement behavior does not match the concept of fault in the principle of fault liability, and as the weaker party, network users often have difficulty in providing evidence. Even if the principle of presumed fault can alleviate the difficulty of providing evidence, the fundamental issue of "whether there is fault or not" has not been solved. The determination of the provider's fault is contrary to the objective fact that the provider cannot foresee and control the generation of infringing content. Therefore, it is more appropriate to adopt the principle of no-fault liability for the providers of Generative Artificial Intelligence services in the attribution of copyright infringement liability.