Japan’s Justice Ministry is preparing to issue guidelines that clarify how existing civil rights may protect voice actors, celebrities and other people from having their voices replicated with generative AI without permission.
A government-appointed expert panel broadly approved the draft at its final meeting on Monday, following discussions about potential civil liability for creating highly similar synthetic voices. The ministry plans to publish the final guidelines as early as the beginning of August.
The document is not legally binding and does not create a new statutory right over a person’s voice. Instead, it explains how established protections, including publicity and likeness rights, could apply to AI-generated audio. The guidance is intended to give individuals, courts and online platforms a clearer framework for assessing claims involving unauthorized voice imitation.
Under the draft, a celebrity’s voice may be covered by the right of publicity, which protects the commercial value associated with a prominent person’s identity. Voices belonging to people who are not public figures may also receive protection under likeness-related rights that restrict unauthorized exploitation of identifiable personal characteristics.
The proposed guidance indicates that making an AI-generated song in another person’s voice could infringe that person’s rights even when the creator discloses that AI was used. A lack of commercial intent would not necessarily prevent a finding of infringement. Those points address a common uncertainty surrounding synthetic media: whether labeling content as artificial or distributing it without seeking profit is enough to avoid legal responsibility.
Intellectual property lawyer Masaya Seki said the guidance should make it easier for affected people to establish the basis of an infringement claim. It could also assist them when seeking removal of content from online services, presenting a civil case or assembling supporting evidence.
Voice actress Yuko Sasaki, an executive member of the Japan Actors Union who is known for her work in the anime series Chibi Maruko-chan, welcomed the panel’s decision.
“This is a major step toward protecting our rights,” Sasaki said.
The union has campaigned for safeguards against generative AI since 2023, as increasingly accessible tools made it easier to reproduce recognizable voices. Sasaki said some performers had felt compelled to tolerate unauthorized uses because responding publicly could itself affect their reputation. She expressed hope that clearer guidance would discourage such conduct.
The draft also addresses sexually explicit deepfakes made with a person’s image or voice. It says such material damages personal dignity and is highly likely to infringe portrait and related rights. That interpretation could help victims pursue civil claims even though the guidance itself carries no binding legal force.
Sumire Nagamori, who leads Hiiragi Net, a private organization that monitors child sexual abuse material and other harmful online content, said explicitly identifying such conduct as rights infringement could encourage victims to bring lawsuits. She cautioned, however, that civil litigation remains difficult because of legal costs and the challenge of identifying anonymous creators or distributors.
Yoji Kudo, a lawyer specializing in sexually exploitative imagery, argued that civil remedies alone would not provide sufficient deterrence. He called for stronger legal rules specifically capable of addressing deepfake pornography.
The guidelines therefore offer a clearer route for removal requests and civil proceedings but do not resolve the practical obstacles facing victims. Their effect will depend in part on how courts apply existing rights to synthetic media and how readily online services respond when people challenge unauthorized AI replicas of their voices.
Sources: Generative AI intellectual property