Look Who’s Talking: Voice Cloning as Tension Point Between Identity and Data

Generative AI has quickly expanded voice cloning’s capabilities, while also lowering barriers for using this technology. Questions about how voice cloning may impact our claim to ownership over our voices remain open, with legal standards lagging in protections for everyday people. This suggests there is a critical need for researchers to develop more frameworks to understand the complexities of voice and voice data as emerging sites where extraction and exploitation can occur. To this end, we explore voice cloning as a tension point that bridges the competing goals of how voice functions in social versus commodity frameworks. We approach this topic through two main research questions: (1) What do different paradigmatic understandings of voice reveal about the power dynamics of voice cloning? (2) How is voice characterized in legal cases and discourse around voice cloning? Using lawsuits from the United States (US) as evidence, we investigate how different voice paradigms are materialized in court decisions and argumentation. We identify key cases where voice cloning is litigated as misappropriations of identity, demonstrating the incompatibility between the value of voice as an intrinsic property of the self and the value of voice as a product. Our framework provides theoretical footholds for describing the ontological complexities of voice cloning and suggests new paths forward for researchers and policymakers.