Skip to content
Rush Commerce
AI & Automation3 min read

Tokyo court: AI voice clones can breach publicity rights

Japan's first AI voice clone ruling says a voice is a publicity right. If your ads or voice agents use a cloned or sound-alike voice, get a signed license.

A Tokyo court just ruled that a human voice is protected as a publicity right, in Japan's first lawsuit over AI voice cloning. The plaintiff did not get the order he asked for. That is not the part to remember. The part to remember is the test the court set: if a voice has commercial appeal and someone uses a copy of it to cash in on that appeal, that use can infringe. Every business that runs AI voiceover in ads or puts a "sounds like" voice on its phone agent now has a clear line in the case law.

What actually happened

On September 30, the Tokyo District Court ruled on a suit from voice actor Kenjiro Tsuda, known for Jujutsu Kaisen. Nikkei Asia reports that an anonymous account posted 188 TikTok videos between July 2024 and September 2025, all narrated with AI speech that resembled Tsuda's voice. The videos covered urban myths and trivia. Tsuda's side said the account made 500,000 to 750,000 yen a month.

Tsuda sued TikTok's operator to get the videos taken down. The operator argued the narration was a "standard male voice." The poster said the model was trained on a friend's voice. By June, the account was gone.

Presiding Judge Aya Takahashi dismissed the takedown request, because nothing was left to take down. But per Jiji Press via Nippon.com, the court held that a voice, like a likeness, is a symbol of a person's identity and falls under the right of publicity. Unauthorized use of a performer's voice can infringe when the voice has commercial appeal and is used to exploit it. The court did not decide whether Tsuda's own rights were breached.

Why AI voice cloning rules matter for your business

This is a Japanese ruling. It does not bind a US court. But it lines up with where US states already are, and it removes the "it's only a sound-alike" defense from the conversation. Note the two excuses that failed to end the case: "it's a generic voice" and "we trained it on someone else."

What we tell clients:

Every voice gets a paper trail. For any synthetic voice in an ad, IVR or voice agent, keep a signed license from the person whose voice it is, or the vendor's written terms for a stock voice. Store it next to the asset.

No "make it sound like" prompts. Do not ask a TTS tool or a freelancer for a voice "like" a known actor or creator. That prompt is the evidence.

Log the voice ID. Record which voice model and version each campaign and agent uses. When a complaint comes in, you can answer in an hour, not a week.

Have a pull switch. Your voice agent should be able to swap voices with a config change. If a claim lands, you change the voice today and argue later.

Key takeaways

  • Tokyo District Court ruled on Sept 30 that a voice is protected under the right of publicity
  • The test: a voice with commercial appeal, used to exploit that appeal, can infringe
  • "Generic voice" and "trained on someone else" did not end the case
  • Keep a signed license or vendor terms for every synthetic voice you ship
  • Build voice agents so the voice is a swappable config value

A voice agent you can't re-voice in an afternoon is a legal risk with a phone number. We build voice and IVR systems with licensed voices, logged voice IDs and a one-line swap. See how we build them, or tell us what your phones need to do.

Sources: Nikkei Asia, Jiji Press via Nippon.com.

  • #ai-voice-cloning
  • #publicity-rights
  • #voice-agents
  • #ai-compliance
  • #japan
TR

Tommy Rush — Founder, Rush Commerce

Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More

Get The Rush Report weekly — one email, zero fluff.