Legal Risk Determination of “AI Face Swapping”

Abstract

With the rapid development of artificial intelligence technology, AI face-swapping technology has been widely applied in fields such as film, television, and advertising. However, it has also triggered numerous legal risks. Therefore, by analyzing relevant legal provisions and typical cases, clarifying the criteria for determining copyright and portrait right infringements in AI face-swapping activities, as well as the principles for allocating liability among multiple subjects, effective legal guidance can be provided for the compliant application of AI face-swapping technology and for rights holders to protect their rights. At the copyright level, the determination of the “access” element should adopt a relaxed standard, and AI face swapping should be assessed for substantial similarity through a combination of the “overall impression method” and the “abstraction-filtration-comparison method.” At the portrait right level, as long as the public can associate the synthetic content with a specific natural person, it constitutes “identifiability” and is suspected of infringing personal information, regardless of whether there is profit motive. In practice, the infringer must simultaneously bear joint liability for both copyright and portrait right infringements; if a platform fails to promptly fulfill its “reasonable care” obligation by taking necessary measures such as deletion or blocking, it shall bear joint and several liability; and if the provider of AI face-swapping technology fails to establish necessary mechanisms to fulfill its reasonable control obligations, it shall bear presumed fault liability.