Generative artificial intelligence (AI) has taken nearly every industry by storm, and the global fashion industry has not been spared. The technology is already having an impact on one of the more human aspects of the industry\u2014modeling. For instance, Maison Valentino announced a new collection that was presented entirely with AI-generated models. Potentially more troubling is the experience of Shereen Wu, a young Taiwanese American model who recently walked for Michael Costello and had her face replaced with \u201cthe face of a White woman she didn\u2019t recognize.\u201d Wu walked in Costello\u2019s show at the Art Hearts Fashion Show in October 2023. In the days following the show, according to Wu, the designer posted a video of the show to his Instagram account; yet instead of the video showing Wu, her face had been replaced, presumably through the use of a generative AI editing tool. In a statement following the post, a representative stated that \u201cneither [Costello] nor [their] team was responsible for such alteration,\u201d but did not explain how the altered photos were ultimately posted to the designer\u2019s Instagram. Unfortunately, Wu\u2019s experience is not likely to be unique. For a field composed largely of women, and with women of color continuing to be underrepresented, Wu\u2019s experience serves as a warning for what may be to come\u2014especially considering the existing constraints and concerns with modeling contracts.<\/p>\n
Valentino\u2019s collection, Wu\u2019s experience, and AI\u2019s potential to \u201celiminate the associated [employment and labor] challenges and expenses\u201d associated with human models, raises questions of how models, especially models of color, might be able to protect themselves in this shifting industry. These questions are front in center when considering a hypothetical runway show, occurring in a completely virtual space (e.g., the \u201cMetaverse\u201d). To develop the virtual representations of models, a company will need to rely\u00a0 on existing material to create a virtual model and their walk. To accomplish this virtual recreation, the developing company may need to either reference, or reproduce, a human model\u2019s walk or the company may \u201ctrain\u201d the AI model using footage and imagery from a human model\u2019s walk.This Essay seeks to explore this hypothetical under the theory that if copyright law can be wielded by models, particularly their walks, then fashion companies may be more reticent to train their generative AI technologies using video of models, as such training would then infringe a model\u2019s intellectual property.<\/p>\n
Part I begins with a brief landscape analysis of the Copyright Act and what additional information can be gleaned from existing regulatory and guidance documents. Building upon this, Part II examines two Ninth Circuit cases in which the court disposed of plaintiffs\u2019 attempts to rely on the Copyright Act\u2019s enumerated \u201cchoreography\u201d category. In Part III, this Essay applies the statute and lessons learned from the Ninth Circuit to theorize how a model might argue that their walk fits within the umbrella of the Copyright Act. Finally, Part IV offers a few thoughts in conclusion, including how models may be able to leverage state right of publicity laws given the seemingly closed federal pathways.<\/p>\n