Japan Actors Union Issues AI Voice Guidelines: Mandating Full-Process Consent and Performance Royalties

Last Thursday, the Japan Actors Union (JAU), representing approximately 2,600 members, released the first version of its generative AI usage guidelines. This marks one of Japan's earliest industry-specific norms targeting AI voice recording, signaling a shift from individual protests to collective rule-making in the face of technological disruption.

According to reports by Automaton, the core requirement of the guidelines is "consent at every step." Any entity wishing to replicate an actor's voice must obtain the individual's permission throughout the entire process; AI companies planning to use actors' voices for model training must also secure prior authorization. Furthermore, the guidelines stipulate that AI-generated speech used with authorization should be classified as a form of "performance," requiring companies to pay actors fees or royalties. This clause attempts to establish an economic return mechanism through industry contracts, given the current gaps in copyright law regarding AI voice protection.

Japanese voice actors have long struggled with the unauthorized appropriation of their voices. In 2024, a group of voice actors formed an organization to protest AI misuse, highlighting legal deficiencies. Prominent voice actor Kenjiro Tsuda even sued TikTok over videos extensively using his voice, though the case was dismissed after the videos were removed, illustrating the difficulties of enforcement. At the press conference, Masako Nozawa, famous for voicing Goku in "Dragon Ball," told NHK: "Using my voice without my permission... is no different from theft." She emphasized that even minimal use requires prior consent and criticized AI-replicated voices as lacking emotion.