AI Voice Cloning Law: What Japan's Ruling Means for Brands

Yes, Japan now protects a person's voice from unauthorized AI imitation. On September 30, 2026, the Tokyo District Court held that a voice is a symbol of personality and can be protected under Japan's right of publicity, the first Japanese judgment to go beyond name and likeness, according to law firm Orrick's analysis of the Tokyo ruling. The court still dismissed the case, so the principle is set but the test for infringement is not.

The dispute involved an anonymous account with more than 210,000 followers that posted at least 188 AI-narrated videos between July 2024 and September 2025. Viewers said the narration sounded like a well-known voice actor. Any brand that runs an AI voiceover resembling a known performer now faces a stated legal theory, even though no court has applied it to a campaign yet.

What did the Tokyo court decide about voice rights?

The Tokyo District Court decided that a person's voice can be protected by the right of publicity in the same way as a name or a face. The judge applied the test from the Supreme Court's 2012 Pink Lady decision, under which use of a voice infringes when it is aimed solely at exploiting the voice's power to attract customers.

That test is narrow. A monetized narration channel mixes entertainment and profit, and the court did not say whether that mix meets it. A Nagashima Ohno & Tsunematsu summary of the ministry report words the test as use "mainly" for that purpose, so the threshold may shift in translation.

Why did the court dismiss the claim?

The Tokyo District Court dismissed the claim because the videos and the account were already deleted before judgment. The poster closed the account in May 2026, and the platform operator had removed the videos from its servers. With nothing left to take down, the court found the platform had no further duty to remove content.

The court did not rule on whether the 188 videos infringed anything. It also skipped the plaintiff's unfair competition argument, which cited monthly poster earnings of JPY 500,000 to 750,000. Orrick reads the result as a sign that fast takedown can protect a platform.

How does Japan's Ministry of Justice report fit in?

Japan's Ministry of Justice published a study group report on August 7, 2026 that reaches the same conclusion on voice as the court did 54 days later. Jiji Press reported that the report treats voice as a symbol of personality protected by the right of publicity. The report has no binding force, but it works through seven hypothetical cases on when AI use infringes.

What does the ruling mean for brands, agencies and talent?

The ruling raises legal risk for four groups: brands, agencies, talent representatives and platforms. The table pairs each with its risk and the action that follows.

Illustrative scenario: a consumer brand hires a vendor to produce a holiday radio spot with an AI narrator. Listeners post that the narrator sounds like a famous audiobook reader. The brand has no record of what trained the model, so it cannot answer the performer's lawyer.

How does Tennessee's ELVIS Act compare?

Tennessee already protects voice, including AI simulations, under its ELVIS Act, which took effect July 1, 2024. Manatt's summary of the ELVIS Act says the law covers voices whether the sound is the real voice or a simulation, and that it reaches advertising uses.

The Tennessee law also extends liability to owners of advertising media who knew, or reasonably should have known, about an unauthorized use of a voice.

What did the ruling leave open?

The ruling leaves five questions open that decide whether a brand's AI voice is safe to run.

• How closely a synthetic voice must match a performer's voice to count as that voice, since the court set no evidence standard.
• Whether a platform must remove infringing content that is still live, and when hosting it creates liability.
• How the sole-purpose test applies to monetized content that also entertains.
• Whether voice cloning can be challenged under Japan's Unfair Competition Prevention Act.
• Whether Japan passes a statute. The Japan Actors Union is calling for one, and an NPO survey cited by Orrick found over 40,000 social media posts suspected of using celebrity voices and likenesses without permission, with an estimated impact of JPY 2 billion to 4.5 billion.

What should a PR team check before running synthetic audio?

A PR team should run four checks before any campaign uses synthetic audio, and keep the results on file. The first three are EPR's working screen, not a legal standard, because the Tokyo court set no similarity test.

1. Name the voice source. Ask the vendor whose recordings trained the model and get the answer in writing.
2. Test for resemblance. Play the audio for five people outside the project and ask whom it sounds like. If two or more name the same performer, stop.
3. License or replace. If a named performer comes up, get a written license or change the voice.
4. Log everything. Keep the vendor's answer, the test results and any license in the campaign file so a takedown or dispute can be answered in a day.

Legal reviewers can start with EPR's guide to AI tools for legal communications teams. Synthetic media is also entering paid campaigns, as 2026 midterm campaign communications data on AI video shows.

What is the decision for communications teams?

License the voice or replace it. The Tokyo court punished no one, but it set the principle, and the Ministry of Justice report points the same way. Brands that log their voice sources now will answer a claim faster than brands that start after one arrives.