Voice rights have their first court ruling in Japan. On 30 September 2026, the Tokyo District Court said that a person’s voice, like their likeness, is “a symbol of individual personality”, and that using a performer’s voice without permission can infringe publicity rights when the use aims to exploit its commercial pull. The case was brought by Kenjiro Tsuda, the voice of Kento Nanami in Jujutsu Kaisen, against TikTok’s operator over AI-narrated videos that sounded like him.
Yet Tsuda lost. The court dismissed his demand that TikTok delete the videos, because the anonymous account behind them had already vanished and TikTok’s servers had purged the data in June. Both sides now say they won, and both are partly right.
We covered the case before the verdict in our explainer on AI voice cloning and the Tsuda lawsuit. This follow-up reads the nine-page judgment itself, explains what the court decided and what it left open, compares Japan’s new position on voice rights with other countries, and sets out what brands, creators and platforms should do now.
Table of contents
- What the Tokyo Court Decided on Voice Rights
- Inside the Nine-Page Voice Rights Judgment
- How the Voice Rights Test Borrows From Pink Lady
- Timeline: From TikTok Videos to a Voice Rights Precedent
- Why Both Sides Claim Victory on Voice Rights
- The Questions the Voice Rights Ruling Leaves Open
- How Japan’s Voice Rights Compare Abroad
- What Voice Rights Mean for Brands, Creators and Platforms
- Voice Rights FAQ
- References and Further Reading
What the Tokyo Court Decided on Voice Rights
The judgment, published on the courts’ own database, is short. Its key passage is a single sentence in the court’s reasoning.
The sentence that matters
Translated, the court said: “A person’s voice can be described, like a person’s likeness, as a symbol of individual personality. The unauthorised use of, for example, a performer’s voice is properly understood to infringe publicity rights where the voice has customer-attracting power that promotes the sale of goods, and the use is aimed exclusively at exploiting that customer-attracting power.” Associated Press reporting rendered the first part as “The human voice is symbolic of individual personality, just like one’s portrait.”
Why Tsuda still lost
Straight after that sentence, the court turned to whether the videos still existed. It found that the poster deleted the account on 29 May 2026, did not apply to restore it within 30 days, and that the account’s data was deleted from TikTok’s servers at the end of 28 June 2026. Tsuda’s lawyers argued that services often keep data longer than they say, but the court accepted TikTok’s internal rules and published policy as consistent with the deletion.
What the court did not decide
Because there was nothing left to delete, the court said it would rule on no further issues. It did not decide whether the AI narration actually infringed Tsuda’s voice rights, whether his voice qualified for protection under the Unfair Competition Prevention Act, or whether TikTok, as the platform, could be made to remove such videos. The formal result was simple: “The plaintiff’s claims are all dismissed,” with Tsuda bearing the court costs.
Inside the Nine-Page Voice Rights Judgment
The judgment adds detail that most news reports skipped. The table sets out the basic facts of the case.
| Item | Detail |
|---|---|
| Court | Tokyo District Court, Civil Division 46 |
| Case | Reiwa 7 (wa) No. 70635, a claim to delete posted articles |
| Bench | Presiding Judge Aya Takahashi and two associate judges |
| Parties | Kenjiro Tsuda v TikTok Pte. Ltd., a Singapore company |
| Legal bases | Unfair Competition Prevention Act, Article 3, or publicity rights |
| Oral argument closed | 3 July 2026 |
| Judgment | 30 September 2026: claims dismissed, costs to the plaintiff |
The account was named 八海, not Nanami
AP described the TikTok account as “Nanami”. The judgment gives its name as 八海, which means “eight seas”. Tsuda’s character is Nanami, written 七海, “seven seas”. The account name was one numeral away from the character, a detail that sits alongside a profile picture AFP said resembled a Jujutsu Kaisen character voiced by Tsuda.
The friend’s-voice disclaimer
The account’s front page linked to an outside site saying the narration was not Tsuda’s voice fed into an AI, but data from an AI trained on “a friend who is good at voice impressions”. TikTok relied on that statement to argue there could be no confusion. Viewers were less convinced. The judgment quotes comments such as “I can hear Tsuken’s voice” and “they say it’s a friend who’s good at impressions converted with AI, but that’s a lie, right?”
The size of the audience
According to the judgment, the account posted 188 videos between July 2024 and September 2025, mixing images and subtitles with trivia narration on urban legends and the paranormal. It had more than 210,000 followers in November 2025, and each video drew thousands to tens of thousands of likes. Tsuda’s side cited the Jujutsu Kaisen 0 film’s audience of 9.8 million as evidence of how well known his voice is.
How the Voice Rights Test Borrows From Pink Lady
Japan has no statute that creates publicity rights. They are judge-made, and the leading case is the Supreme Court’s Pink Lady decision of 2 February 2012, about photographs of the pop duo used in a magazine diet feature. The Tokyo court’s voice rights sentence copies that decision’s structure almost word for word.
Three ways to exploit a famous identity
Pink Lady said that using someone’s name or likeness infringes publicity rights only when the use aims exclusively at exploiting its customer-attracting power, giving three typical examples. Tsuda’s lawyers argued the TikTok videos fitted all three; TikTok disagreed with each.
| Pink Lady category | Tsuda’s argument | TikTok’s reply |
|---|---|---|
| The identity is the product people enjoy | The videos used his voice as something to be enjoyed for itself | The narration was not the object of appreciation |
| Used to differentiate goods | Ordinary trivia gained value only from the voice | Views did not grow because of the voice |
| Used as advertising | The voice pulled viewers to commonplace content | Not advertising for any product |
“Exclusively” is a high bar
The word doing the work is “exclusively”. A news report, a review or a parody that happens to use a famous voice is unlikely to qualify, because its purpose is not solely to trade on that voice’s pull. That limit protects expression, but it also means voice rights will be argued case by case.
The guideline had already said it
Tatsuhiro Ueno, an intellectual property professor at Waseda University, told Bengo4.com that the ruling was not a surprise in substance. The official commentary on Pink Lady by a Supreme Court judicial research official had read the “etc.” in “name, likeness, etc.” as covering voice, and the Justice Ministry’s August guideline said so explicitly. What is new, he said, is that “a Japanese court has stated it clearly for the first time”.
What the ministry guideline adds
The Justice Ministry’s report, finalised on 7 August, goes further than the court on two points. It says a voice actor’s voice can carry customer-attracting power on its own, even if the person’s name and face are not widely known. It also says deletion can be sought from the platform operator, not only from the person who posted. The guideline is not binding, but courts and companies read it, and it fills gaps the judgment left open on voice rights.
Why voice actors are a special case
Many voice actors are famous by ear rather than by face. A fan may recognise Tsuda’s voice in seconds without knowing what he looks like. A test built for photographs and names had to stretch to cover that kind of fame, and the ministry’s footnote on unknown faces shows that voice rights were drafted with exactly these performers in mind.
Timeline: From TikTok Videos to a Voice Rights Precedent
The case ran in parallel with the government’s own work on AI and voice. The dates below come from the judgment, Japanese press reports and the Justice Ministry.
| Date | Event |
|---|---|
| July 2024 | The account begins posting AI-narrated videos |
| August 2025 | Court orders disclosure of sender information |
| September 2025 | Last of 188 videos; Tsuda asks TikTok to stop transmission |
| November 2025 | Tsuda files the deletion suit |
| 29 May 2026 | The poster deletes the account |
| 28 June 2026 | Account data purged from TikTok’s servers |
| 3 July 2026 | Oral argument closes |
| 7 August 2026 | Justice Ministry publishes its voice and likeness report |
| 30 September 2026 | Judgment: voice protected in principle, claim dismissed |
The gaps that shaped the outcome
Counting the days shows why the case ended as it did. The account vanished 30 days before its data was purged (29 May to 28 June), the court then took 89 days from the close of argument to judgment (3 July to 30 September), and the ministry’s report landed 54 days before the ruling (7 August to 30 September).
Days between key events in the Tsuda case (2026)
What the account earned
Tsuda’s lawyers estimated, from TikTok’s view-based creator payouts and an average of about 1.47 million views per video, that the poster earned ¥500,000 to ¥750,000 a month. Across the 15 months from July 2024 to September 2025, that range multiplies out to ¥7.5 million to ¥11.25 million, if the monthly rate held throughout. The court made no finding on the figure.
Implied earnings over 15 months, millions of yen (lawyers’ monthly estimate x 15)
Why Both Sides Claim Victory on Voice Rights
The ruling “ended up making both sides look victorious,” Tsuda’s lawyer Takashi Hirano told AFP. With the account gone, the lawsuit had achieved what it set out to do.
Tsuda will not appeal
Tsuda never sought damages, only deletion, and his side has said it will not appeal. Professor Ueno noted the cost of that choice: without a damages claim, the court was never forced to decide whether this particular voice was infringed. He called that “a little regrettable”.
TikTok’s response
TikTok told AFP: “We will continue to sincerely work on the balance between technology development and rights protection,” adding that it would study the judgment before deciding its next step. In court it had argued that the narration was a “generic male voice” and that it merely provides a place to post.
The actors’ union
The Japan Actors Union, which backed the suit, posted that it supported Tsuda and was grateful for “his courage”, and said it would comment further after reading the full judgment. Voice actors have campaigned since October 2024 under the banner “NO MORE unauthorised generative AI“.
The platform effect
Lawyer Kensaku Fukui told TV Asahi’s morning show that suing the platform, rather than only the anonymous poster, was the important move. In his view, platforms now find it harder to ignore notices asking them to remove unauthorised uses of a voice.
The Questions the Voice Rights Ruling Leaves Open
A general statement of principle is a start. Several practical questions about voice rights and AI remain unanswered.
How similar is too similar?
Ueno pointed out that proving a cloned voice is the same as, or similar to, the real one is hard, and the technical methods for doing so are unsettled. Tsuda’s team commissioned an acoustic comparison; the court never reached it.
How a soundalike is made
Modern voice cloning pairs speech synthesis with natural language processing: one model learns the timbre and rhythm of a voice from recordings, and another turns any script into speech in that voice. Commercial tools advertise cloning from short samples. That is why the source of the training audio is so hard to pin down, and why the court’s focus on commercial pull, rather than technique, matters for voice rights.
What if the model was trained on someone else?
The poster claimed the model learned from a friend’s impression, not from Tsuda’s recordings. The court did not treat that claim as fact or decide whether it matters. The test it stated looks at whose customer-attracting power is exploited, not at what the training data was, which suggests a soundalike cannot hide behind its training set.
What about impressionists?
On the same TV Asahi programme, lawyer Toki Yuki drew a line for comedians who do impressions. If audiences come for the impressionist’s own skill, there is no problem. If the act trades on the original performer’s pull without permission, he said, an infringement “is possible”.
Who is responsible: the poster or the platform?
Ueno listed this as a key gap. The judgment did not decide whether a platform operator, separate from the poster, can be required to act. The Justice Ministry’s guideline suggests deletion can be sought from the platform; a court has yet to say so.
How long do voice rights last?
Publicity rights after death, and how long protection lasts, remain open in Japan, Ueno added. For voices that AI can reproduce indefinitely, that question will not stay theoretical for long.
How Japan's Voice Rights Compare Abroad
Japan has now joined a small group of jurisdictions where a voice has explicit legal protection. The approaches differ widely, as the table shows.
| Jurisdiction | Source of protection | Status |
|---|---|---|
| Japan | Judge-made publicity rights | Voice confirmed by a court, 30 Sep 2026 |
| Tennessee, US | ELVIS Act | Signed March 2024; voice added to publicity rights |
| US federal | NO FAKES Act | Cleared Senate Judiciary 18 June 2026; floor vote blocked 30 Sep |
| China | Civil Code personality rights | Beijing Internet Court AI voice ruling, April 2024 |
| India | Personality rights in the courts | Bombay High Court order for singer Arijit Singh, July 2024 |
| United Kingdom | No personality right; passing off | False endorsement since Irvine v Talksport (2002) |
The same-day US setback
On the day of the Tokyo ruling, Senator Marsha Blackburn tried to pass the NO FAKES Act by unanimous consent before the Senate left for the midterms. Senator Ted Cruz objected, saying the bill’s exceptions for satire and political commentary were not clear enough, Music Business Worldwide reported. The bill remains on the Senate calendar.
Where the UK stands
The UK has no general right to one’s image or voice. A celebrity whose voice is cloned for an advert would usually rely on passing off, which requires showing a false suggestion of endorsement, or on data protection law where recordings are personal data. That leaves a gap for the kind of anonymous, ad-funded channel in the Tsuda case.
What Voice Rights Mean for Brands, Creators and Platforms
The ruling applies in Japan, but its logic travels. Any business that uses synthetic voices, or hosts content that does, should treat famous voices as protected property rather than a free style.
Brands and agencies
Do not brief a “sounds like” voiceover of a known actor or presenter for an advert, explainer or social campaign aimed at Japan. Use licensed voices with written consent, or synthetic voices that are not modelled on an identifiable person. Keep the consent and the voice model’s provenance on file. Our compliance team can help set that process up.
Creators
Under TikTok’s own rules, AI-generated audio that mimics a real person’s voice must be disclosed, while generic text-to-speech narration is exempt. A disclaimer saying a soundalike came from “a friend” did not stop viewers recognising Tsuda, and it would not obviously answer a voice rights claim either.
Platforms
Japan’s Information Distribution Platform Act, in force since April 2025, already requires large platforms, including TikTok, to handle removal requests quickly. After this ruling, a notice that a channel exploits a famous voice deserves the same urgency as a likeness complaint. Platforms should also preserve evidence when an account disappears mid-dispute; the Tsuda case turned on deleted data.
UK companies selling into Japan
Treat Japanese voice rights as you would trade mark clearance. Check scripts, voice models and influencer content before launch, and build a takedown route that can act within days. The same audit will serve you if the UK or the EU legislates.
Voice Rights FAQ
Did Kenjiro Tsuda win his case?
Not formally. The court dismissed his claim because the videos had already been deleted. But it stated, for the first time in a Japanese court, that a voice can be protected by publicity rights.
What test did the court set for voice rights?
Using a performer’s voice without permission infringes publicity rights where the voice has customer-attracting power and the use aims exclusively at exploiting it. That mirrors the Supreme Court’s 2012 Pink Lady test for names and likenesses.
Did the court decide that the AI voice copied Tsuda?
No. Because the data had been deleted, the court expressly declined to decide the remaining issues, including whether this narration infringed his voice rights.
Is AI voice cloning now illegal in Japan?
No. Cloning a voice is not banned. Using a recognisable voice without permission, mainly to trade on its appeal, can now be challenged in court with this ruling as support.
Will there be an appeal?
Tsuda’s side has said it will not appeal, and he never claimed damages. TikTok said it would study the judgment.
Does the UK protect voices the same way?
No. The UK has no standalone personality right, so cloned voices are usually challenged through passing off or data protection law.
References and Further Reading
Judgment, Tokyo District Court, Reiwa 7 (wa) No. 70635 (Courts in Japan, PDF)
Japan court protects human voices in landmark AI case (AP)
Japan court rules voice rights protected in AI cloning case (Anadolu Agency)
Interview with Professor Tatsuhiro Ueno on the Tsuda judgment (Bengo4.com, Japanese)
Voice rights ruling and impressionists (J-CAST News, Japanese)
Report on civil liability for unauthorised use of likeness and voice (Ministry of Justice, Japanese)
Japan Actors Union statement on the Tsuda ruling (Sponichi, Japanese)
More AI coverage: explore Progressive Robot's AI Models, Tools & Releases hub — hands-on reviews, setup guides and benchmarks in one place.
