Zizi in Motion - A Deepfake Drag Utopia (Movement by WET MESS)

In this video installation, artist Jake Elwes disrupts artificial intelligence (AI) systems by employing drag—a performance form that challenges societal norms around gender expression. Elwes collaborated with London drag queens, kings and things to construct deepfakes— AI-generated versions of themselves that can imitate real movement. Elwes first filmed each performer from different angles, and then used that footage to train AI models from scratch to animate their likenesses. It was important to the artist not to use stolen images and for the performers to consent to the use of their likenesses and retain control over their data. As the simulations of the cast move to drag legend WET MESS’s choreography, their wigs, makeup, and costumes become distorted. AI’s failure to accurately render the drag performers suggests the gender bias and insufficient queer representation in training datasets for AI models. By engaging with the system’s limitations, Elwes critiques AI’s normative framework and addresses ethical concerns around data collection. They also explore how expressions of queerness can move beyond binary labels that classify humans as male or female. Drag becomes a playful tactic for the artist to demystify and subvert these often exploitative technologies and create artwork by and for their community.