Spike News

Japanese Voice Actors Push Back Against Unauthorized AI Voices

According to Japan Broadcasting Corporation (NHK), cases of generative AI illegally using others' voices are occurring frequently. In this context, organizations representing Japanese entertainment professionals such as voice actors held a press conference on the 15th, announcing the creation of the first guidelines regarding the use of AI in voice processing.

In the process of advancing this initiative to protect 'voices', Ken Yokozawa, a leading figure who has shaped Japan's voice actor culture, spoke in an interview for the first time. Regarding the issue of using AI voices, she said: 'I cannot tolerate the unauthorized use of others' voices without permission. I hope everyone will follow the rules.'

Yozawa Masako once voiced the protagonist Sun Wukong in the popular anime series "Dragon Ball". At 89 years old, she is still active in the industry.

Noyez Yasuko said that she was not very aware of her voice being misappropriated by AI until then.

Japanese Voice Actors Push Back Against Unauthorized AI Voices

Report screenshots

After watching the AI-synthesized video again, she said, "This is not my voice. It was used without any acknowledgment from me. I simply cannot forgive this; it's nothing short of theft. I'm very angry about this. I believe that the voice belongs to me, and so does the performance. Even if it's only used in small amounts, permission should be obtained beforehand."

She also said, “The voices generated by AI may sound very much like me, and can even be quite realistic. But there is no heart in them. Without heart, it doesn’t work. No matter what role it is, even if it’s just a line about crying, if you cry with true emotion, the result will be completely different.”

Recently, there have been a growing number of cases where Japanese generative AI systems haveappropriately plagiarized the voices of voice actors and other celebrities. Notably, a type of video is widely shared on social media platforms: these videos use the voice talents of voice actors to perform in popular animated characters, making the characters sing lines that they have never actually spoken.

Among them, the well-known voice actor Tsuda Kenjiro filed a lawsuit against the operating company of the video platform TikTok regarding the issue of “videos using generative AI to impersonate my voice and upload them to the internet”. He demanded that the relevant videos be deleted.

In a judgment issued on the 30th of last month, the Tokyo District Court ruled for the first time that the unauthorized use of another person's voice infringes upon a celebrity's right to prohibit the commercialization of their likeness and other assets that are used without authorization (right of publicity).

Facing the chaos of generative AI stealing voices, the Japanese Haiku Actor Association held a press conference in Tokyo. The organization has about 2,600 members, including actors and performers.

This newly established group guidelines stipulate that if the voices of voice actors and others are used for AI training or AI-generated content, it is necessary to inform the individuals themselves or their agency of the purpose of use, and obtain their consent in advance.

Additionally, if using AI to translate my voice into another language, permission must be obtained for each language separately. When training and generating with the voice of a deceased person, it is necessary to respect the deceased’s wishes during their lifetime, as well as obtain the consent of the survivors.

Once violations occur, the right holder can demand the immediate cessation of the dissemination and sale of AI voices. The claim can be traced back to the date when the unauthorized use began. The right holder can also demand compensation equal to the sales revenue from the unauthorized activities.

Japanese haiku association plan for disseminating guidelines to related businesses.

Attending the press conference, voice actor Masato Date (known for works like "Cherry", and character Yuzo Grandpa) said, "Fabricating lines I have never spoken and songs I have never sung, while also making profits from this, is something I simply cannot accept. What worries me most is that unauthorized AI voices are becoming increasingly common, and in the end, this will destroy the voice acting industry."

"The most core issue is that the jobs of workers in the industry are taken away by AI. We hope to accumulate practical examples based on this guide and eventually push for relevant legislation."

When talking about the creation of this first set of guidelines for the use of AI voices, Yonezawa Masako said, "Rules are being established gradually, and we hope that everyone will abide by them. The premise is that everyone truly adheres to these rules."

She also sent her words of encouragement to the young people who aspire to become voice actors: “Strive to be a voice actor whose voice makes people think of you immediately. If there is no sincerity, then lines are just words read out. I hope you can give these lines a living, vibrant meaning.”

Finally, she recorded a video message as the vice president of the Japan Haiku Actors Association: “The progress of AI is truly astonishing. Things that were once fantasies from childhood are becoming reality, and I am thrilled about this. But when it is misused by others, it can be extremely disappointing. I hope everyone will abide by the rules and create a future where both humans and AI can enjoy themselves.”

The lawyer Kenji Fukui, who specializes in AI and rights issues, commented on this public guide: "This is a significant attempt with quite groundbreaking content. It can have a certain deterrent effect on domestic companies and users. Although there are differences in the legal positioning of voice usage and other copyrighted subjects, this guide will have a profound impact on other industries."

He also pointed out, “At this stage, there is no law that directly prohibits AI’s data training activities. This guideline does not have legal enforcement power, and there are challenges in implementing it in overseas large platforms and AI companies. If we can get them to agree to comply with requirements such as disclosing training data, and jointly promote the establishment of international rules with various countries, it would be of great significance.”