Voice is also an identifying factor: Japan made the first ruling on voice cloning using AI

From July 2024 to September 2025, an anonymous TikTok account in Japan, with about 210,000 followers, posted at least 188 videos about urban legends and mysterious stories. Each video of this account averages about 1.47 million views, bringing in 500,000 to 750,000 yen per month. What they all have in common is the voice: deep, husky, very similar to Kenjiro Tsuda, a voice actor familiar to anime fans as Kento Nanami in Jujutsu Kaisen and Seto Kaiba in Yu-Gi-Oh!. It's just that Tsuda never read those videos, that voice was created by AI.

On September 30, 2026, the Tokyo District Court issued a ruling in the case of Tsuda suing the TikTok operator. In fact, this is the first lawsuit in Japan related to voice cloning using AI. Presiding judge Aya Takahashi concluded that "the human voice is a symbol of an individual's personality, as well as image" and personally I think this sentence is much more important than the nature of the case, because it is the first time people have put their voice on par with their face in the eyes of the law.

Image copyright and why voice is now considered an exception

Japan has a legal concept called “publicity right”. Simply put, this is the right of a celebrity to control how others use their name and image to make money. If a company prints a singer's face on a product's packaging without permission, they violate this right. The Japanese Supreme Court has recognized the right of publication since 2012, in the case of singer duo Pink Lady suing a magazine for using their photos without permission.

Tik Tok's argument: it was a generic male voice and the account also quickly deleted the video

On the contrary, TikTok argued that the voice in the videos was just "a generic male voice". They said the AI ​​model was trained from the voice of a friend of the account holder, not from Tsuda's voice. This argument makes sense given the way AI works. No one has the right to own a "voice type" like deep or husky, but in fact, millions of men have this deep voice. If every deep, husky voice was considered an imitation of Tsuda, the law would become absurd. However, Tsuda's side did not leave it alone. His lawyer presented the results of an acoustic analysis and argued that the voice in the video was a deliberate imitation of his signature "deep and shiny" voice. In addition, the court does not consider a voice to be merely anonymous audio data and that using another person's voice without permission may violate the right of publicity when the purpose is to exploit the commercial appeal of that voice.

TikTok dismissed it as a generic voice, and the account also deleted the video before the Court could judge I think the key point lies in the word "commercial attractiveness". The court did not say that anyone with an accent similar to Tsuda would also be in violation. The court said that when a person intentionally creates a voice similar to Tsuda to attract viewers and make money, they are taking away something that belongs to Tsuda. An account that earns hundreds of thousands of yen every month from a voice that is familiar to anime fans can hardly be said to be a coincidence. There is one detail that makes this ruling more symbolic than real. The news reports I read did not specify the amount of compensation, if any, but what is certain is that the court will deny the request to remove the video related to Tsudua, because the account and video have been removed before, there is nothing left to remove. This detail shows a legal problem that is sure to recur over and over again: infringing accounts can make money for more than a year, then disappear when sued. The court's processing time is often quite slow and it was enough for the defendant to disperse the video and grow another account. The court's ruling only has the value of creating a precedent, but it does not recover the money earned, nor does it prevent the next account from committing similar acts. In addition, Tsuda's lawyer also alleged unfair competition practices under Japan's Anti-Unfair Competition Law. This law was created long before voice-generating AI, so its application to synthetic voices is still an open legal question.

Not only Japan, America also has a similar trend and what impact does this have?

In fact, if we take a moment to summarize, we can see that the Tokyo ruling is part of a much broader trend. In August 2026, an expert panel of the Japanese Ministry of Justice recommended treating voices like images. The Tokyo District Court has now moved in that direction even though there is no new law yet.

Scarlett Johansson also forced OpenAI to stop Sky's voice because it surprisingly resembled her The common point of these cases is that the law is gradually accepting an idea: the voice is not just a sound, but a part of identity that others can hear, recognize, and therefore can be exploited for personal or economic gain. In fact, rulings like this do not prohibit AI from creating accents at all. Tools that read text with synthetic voices, self-designed virtual voices or licensed voices are still completely legal. What is being targeted is the deliberate copying of a specific, recognizable person's voice, in order to monetize their appeal. Of course, the boundaries of this will still be controversial with many questions: How similar is it to be considered imitation? Is acoustic analysis enough evidence? What if the model is trained on someone else's voice but the result resembles a famous person? The Tsuda case has not answered all of these questions, and I think the next cases will have to gradually clarify these factors. For the Japanese dubbing industry, an industry where reputation is closely linked to voice quality, this is the first time they have a verdict to rely on. Voice actor Megumi Ogata, who voices Shinji Ikari in Evangelion, also called on Japan to protect voices from AI. For them, the verdict on September 30 is an important step: from now on, their voices are recognized by the Japanese court as belonging to them.