Dismissed but "Voice Is a Right" | The First Ruling in Tsuda Kenjiro's AI Voice Lawsuit
機械翻訳 / Machine-translated
A
AI Friends
1日前・22分で読める
機械翻訳 / Machine-translated
@aifriends
AI Friends（https://aifriends.jp）のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
"Request dismissed" — yet the plaintiff's side was all smiles, saying "our objective was fully achieved." This seemingly contradictory ruling looks set to become a major turning point in an era where AI can impersonate anyone's voice. This article breaks down the substance of the ruling and what it means for voice actors, creators, and everyday social media users.
On September 30, 2026, the Tokyo District Court handed down its ruling in a lawsuit filed by voice actor and actor Tsuda Kenjiro.
The ruling characterized a person's voice as "a symbol of individual personhood, just like a person's likeness."
It further held that when a voice possesses "customer-drawing power" (the ability to attract customers) that promotes the sale of goods and the like, unauthorized use of that voice for the purpose of exploiting that power constitutes an infringement of publicity rights.
Hiroyuki Hirano, the attorney representing the plaintiff, commented: "In the sense that this is a ruling that explicitly places voice within the scope of publicity rights, this is likely the first of its kind."
The focus of the lawsuit was 188 videos posted on TikTok.
An anonymous account had been posting narrated videos using an AI imitation of Tsuda's voice between July 2024 and September 2025. The content centered on urban legends, the occult, and trivia.
The videos averaged approximately 1.47 million views each. The account had amassed more than 210,000 followers, and the poster is reported to have earned 500,000 to 750,000 yen per month in revenue.
The background to the lawsuit is covered in detail in our earlier article, "Tsuda Kenjiro Files Lawsuit Against TikTok Over AI Voice Imitation | The Ripple Effect of 188 Fake Videos".
The reason the court rejected the claim is remarkably simple.
The poster had voluntarily deleted their account on May 29, 2026. Because no reinstatement request was filed within 30 days, the court determined that the video data had been erased from the servers by June 28. In other words, the very content the plaintiff sought to have removed no longer existed at the time of the ruling. The logic behind the dismissal is that a platform operator cannot be ordered to remove videos that are already gone.
Tsuda's side had not sought damages from the outset. Their position was that the case was "not motivated by financial compensation."
Attorney Hirano explained that "since all videos were deleted along with the entire account, the original objective was fully achieved," and confirmed the decision not to appeal.
In practice, even though this ruling is technically a loss on paper, it represents a significant step forward. Because the court explicitly stated in its reasoning that "voices are also protected," it sets a benchmark for future cases of the same kind.
Publicity rights refer to the exclusive right of a public figure to commercially exploit the "customer-drawing power" of their name, face, and similar attributes.
There is no statutory provision for this in Japanese law. It is a right that has been shaped through the accumulation of court decisions.
The landmark case is the Supreme Court ruling of February 2, 2012 (the Pink Lady case). In that ruling, the Supreme Court held that infringement occurs when likenesses and similar attributes are used "exclusively for the purpose of exploiting customer-drawing power."
Until now, however, the protected elements were primarily "name" and "likeness." Whether voice fell within this category had remained unclear.
Have you ever wondered, "Couldn't copyright law protect a voice?"
Copyright law protects recorded audio and performances themselves. The quality of a voice and its characteristic mannerisms are not covered.
An AI-synthesized "sound-alike voice" created from scratch does not copy an original recording. That is precisely why publicity rights have emerged as a crucial alternative avenue of protection.
This ruling is in step with the government's recent actions.
On April 17, 2026, the Ministry of Justice announced the establishment of an expert review panel, which held its inaugural meeting on April 24. At the July 13 meeting, a draft guideline was presented explicitly stating that voices, like likenesses, are entitled to protection under publicity rights.
Then, in August, the report "Interpretive Guidelines on Publicity Rights Infringement Caused by Generative AI" was published. It includes the position that posting AI-generated videos that mimic the voices of voice actors and others on social media for commercial gain may constitute infringement.
However, the guidelines carry no legal binding force. This ruling has now added judicial endorsement to the government's position. For a detailed overview of the review panel's proceedings, see "Ministry of Justice: 'Voice Is a Right' | Path to Damages Claims for Unauthorized AI Audio".
In the United States, rules to protect voices are being established on a state-by-state basis.
In July 2025, a New York state court allowed claims based on the state's publicity rights law to proceed in Lehrman v. Lovo, a case in which two voice actors sued the AI voice company Lovo. The state law prohibits the unauthorized commercial use of a person's "name, likeness, and voice."
At the federal level, the "NO FAKES Act", which would regulate AI reproductions of voices and likenesses, was reintroduced in May 2026. It passed the Senate Judiciary Committee unanimously on June 18 and awaits a full Senate vote.
In August 2026, 80 actors in the UK issued a statement demanding "ownership rights over their voices" (related article).
In the United States, the actors' union SAG-AFTRA has also negotiated with game companies to secure conditions for AI reproduction that include "advance notice, written consent, and payment of compensation."
To summarize: Japan is pursuing voice protection through "case law + administrative guidelines," the United States through "state laws + a federal bill," and the Western industry through "labor-management contracts" — three parallel tracks toward the same goal.
You may have come across TikTok or YouTube videos featuring AI narration that sounds just like a famous voice actor.
This ruling has made it clear that posting such videos for commercial gain can constitute a rights infringement. The excuse of "I only trained it on a friend's voice" will become harder to sustain if listeners perceive the voice as belonging to the actual person.
Imagine a freelance creator who posts trivia videos with an AI voice modeled on a popular voice actor. The more views and ad revenue that come in, the more likely it is to be judged as "exploiting customer-drawing power." Switching to their own voice or a licensed voice now is the safer course of action.
Next, consider an audio director at a game company. Some in the past may have thought, "If we create a voice-actor-style AI voice, we can cut appearance fees." Going forward, obtaining written consent from the actual person and establishing appropriate compensation will be indispensable.
Finally, consider the platforms themselves. This ruling did not directly address the liability of the operating company. Even so, improving the process for handling removal requests and implementing "AI-generated audio" labels are likely to become urgent priorities for every platform.
That said, several issues were left unresolved by this ruling.
First, the standard for similarity — how alike a voice must be to constitute infringement — remains undefined. While the plaintiff presented evidence such as comparisons using multiple deep learning models, the deletion of the videos meant that no concrete threshold was established.
Second, there is the "whack-a-mole" problem of posters disappearing anonymously. Attorney Hirano himself acknowledged that it "tends to become a game of whack-a-mole."
Finally, the extent to which the voices of ordinary, non-famous individuals are protected also remains an open question for the future.
Technically, the claim was dismissed — but the reason was that the videos had already been deleted. The court recognized that voices are protected under publicity rights, and the plaintiff's side, having achieved its objective, has decided not to appeal.
This ruling targets unauthorized use of a voice's customer-drawing power for commercial purposes. Private, non-monetized enjoyment does not immediately become illegal. However, making such content public or monetizing it significantly increases the risk of being found to have committed an infringement.
Publicity rights apply to individuals who possess "customer-drawing power," so they are generally considered not to apply to private individuals. However, the Ministry of Justice guidelines also include voice within the scope of the right "not to have one's voice exploited without permission," leaving open the possibility of protection through a different avenue.
Using voices provided by the service or voices for which rights have been cleared is generally not a problem. If you are creating a voice designed to resemble a specific public figure, it is safest to obtain written consent from the person or their agency in advance.
If you create videos or other content using AI voices, now is a good time to double-check the source of the voices you are using and whether the necessary rights have been cleared.
This article is a cross-post from AI Friends.
