Masako Nozawa condemns unauthorised AI voice clones as Japan sets new rules

Masako Nozawa, the voice of Goku in Dragon Ball, has criticised unauthorised AI copies of performers’ voices as Japan’s actors union publishes its first guidelines for generative AI. The Japan Actors Union (Nippairen) announced the voluntary rules at a press conference on 15 October, calling for consent and payment at every stage of voice-AI use. ¹

What the new guidelines require

The rules ask companies to obtain written permission before training an AI system on a performer’s voice, creating a model and using its output. Consent should spell out the purpose and scope of the use, while buyout contracts are prohibited. The union also says AI-generated voice work should be paid like a new performance, rather than treated as a free extension of an older recording. ¹

The guidance covers more than Japanese-language performances. AI dubbing into another language requires fresh permission and additional payment, according to the union’s announcement. Recreating the voice of a deceased performer also requires permission from their heirs or another rights holder. The rules are designed as a voluntary industry standard, not a law, but Nippairen says it wants them to help close what it describes as a gap in Japanese legal protection for a person’s voice. ¹

Why Masako Nozawa is speaking out

Nozawa, who has played Goku since Dragon Ball first reached television audiences in 1986, sent a video message to the press conference. As reported by Anime Mojo and other outlets, she welcomed the possibilities of AI but said unauthorised use of performers’ voices left her disillusioned. Her comments underline the difference between a licensed voice model and a clone made or published without the performer’s approval. ²

The distinction matters in Nozawa’s own case. Her agency previously worked with CoeFont on licensed AI versions of actors’ voices for uses including assistants, robots and navigation systems, while the union’s new rules address consent, compensation and control across the wider production process. ²

The wider legal question

The union says Japan still lacks a standalone law protecting a person’s voice, and plans to lobby for legal reform. A recent Tokyo District Court ruling involving Jujutsu Kaisen voice actor Kenjiro Tsuda recognised that a person’s voice could fall under publicity rights, according to reporting cited by Anime Mojo, although the court did not order the removal of videos that had already been taken down. ²

Our opinion

The useful part of Nippairen’s announcement is its specificity. Consent at training, model-building and publication gives performers more control than a vague promise to “use AI responsibly”, while payment and language-specific permissions acknowledge that a voice is a continuing performance, not just raw data. The rules are voluntary, so their real test will be whether producers treat them as a baseline before lawmakers catch up.