Japan's Ministry of Justice puts voice inside publicity rights

TL;DR

On 27 July 2026, a Japanese Ministry of Justice study group on civil liability for the unauthorized use of likeness and voice approved its final report, published in August under the subtitle “interpretive guidance on infringement of publicity rights by generative AI.” The report sets out that a person’s voice falls within the “likeness and the like” protected by publicity rights, lays out the test for infringement, and confirms that a takedown can be demanded of the platform operator as well as the poster. The interpretive layer moved forward. What is missing is that every remedy it sets out presumes the generated work already exists — nothing in the framework asks, before generation, whether that use of the voice was authorized against the person’s consent.

What happened

• The study group met five times between 24 April and 27 July 2026, approving its final report at the last session. The report itself is dated August 2026; a summary document was still marked “in preparation” at the time of writing.
• The report takes up directly whether voice falls within “likeness and the like” as used in the Supreme Court’s Pink Lady judgment. It cites the judicial research official’s commentary on that judgment, which treats the phrase as covering information identifying the person — “a signature, an autograph, a voice, a pen name, a stage name, and so on” — and so states plainly that voice is protected by publicity rights. The study group supported that reading, while noting the qualification that what falls within the phrase must be something that can be called a symbol of the individual’s personality.
• On the test for infringement, the report holds that the same three categories that apply to name and likeness apply to voice: using the voice itself as a product to be appreciated in its own right; attaching the voice to a product in order to differentiate it; and using the voice as advertising for a product. Infringement arises where the use can be said to be solely for the purpose of exploiting the customer-drawing power the voice carries. Where a voice actor’s voice itself promotes sales, that power is recognised.
• The report also records that the fit is unsettled for generative AI in particular. Where a video or audio work is made by using someone’s voice without permission through generative AI, the voice becomes embedded in the work, and — in the report’s words — whether that amounts to “attaching” under the second category, or falls under the first or under “and so on,” is not necessarily clear.
• On remedies, the report addresses damages and injunctions for infringement of publicity rights, and holds that for content on the internet a takedown may be demanded “of the platform operator that manages and runs it” as well. An injunction is framed as being “to remove an infringement now being carried out, or to prevent an infringement that would otherwise arise.”
• In the background, a voice actor is reported to have sued the operator of a video platform seeking removal of videos that mimicked their voice with generative AI. This Brief could not confirm that suit from a primary source and so does not go into its particulars. On 28 July 2026, the Japan Actors Union published its position on the guidance, asking that prior consent before generation and platform obligations be stated explicitly in the final text.

The reach of the framework is set by this order.

1. Generative AI produces audio imitating a person’s voice without their consent being checked.
2. A work containing that audio is posted, distributed, and monetised.
3. Only now does a “product” exist, and only now can the customer-drawing-power test be applied to it.
4. The rights holder claims damages and demands removal from the poster or the platform. The work exists throughout those proceedings.

Timeline — disclosure and response

• 2026-04-24: first session of the study group.
• 2026-05-28 / 06-25 / 07-13: sessions two through four.
• 2026-07-27: fifth session. The final report is approved.
• 2026-07-28: the Japan Actors Union publishes its position on the guidance.
• August 2026: the report itself is published, subtitled “interpretive guidance on infringement of publicity rights by generative AI.”

The interpretation, the test, and the account of remedies drawn on here come from the text of the Ministry of Justice’s final report, read directly. The individual lawsuit that has been reported could not be confirmed from either a filing or a statement by the parties, so this Brief does not name anyone or go into post counts or revenue figures. The report is interpretive guidance; it is not legislation, and it creates no new cause of action.

The response after publication:

• The Japan Actors Union welcomed the effort to clarify how the existing frameworks — publicity rights, personality rights, and unfair competition law — reach unauthorized use by generative AI, while asking that the final guidance state prior consent before generation, user responsibility, platform obligations, and protection regardless of commercial intent.
• The report does not treat unfair competition law as a direct subject, but notes its position, on the ground that recognising the difference in scope matters.

Why it wasn’t stopped

The failure here is neither an absence of legal protection nor an interpretation left vague. What was settled was, in every case, what can be done once the work already exists.

The interpretive layer moved forward. Voice was placed within publicity rights, a three-part test was set out, and platform operators were confirmed as parties a takedown can be demanded of. What is missing sits earlier — a step, before generation, that establishes whether that use of the voice was authorized against the person’s consent. The test turns on use “solely for the purpose of exploiting the customer-drawing power the voice carries,” and that purpose can only be judged once a product exists. An injunction, likewise, is framed as removing an infringement now being carried out or preventing one that would otherwise arise: it starts from an infringing act.

Publishing is an action. Without the use of that voice having been checked against consent before that action, a takedown is only an acknowledgement after the imitation has spread.

That the report calls the fit for generative AI “not necessarily clear” bears this out rather than weakening it. Once a voice is embedded in a generated work, which category it belongs to becomes something to argue after the fact. Check consent before generation and the argument does not arise. This is the same shape as Brief 105, where a 100% takedown rate still left a fresh model of the same person being posted, and it runs in the same direction as Brief 117, where consent was confirmed by a single checkbox, and Brief 054, on moving the permission principle ahead of generation.

What proof would have changed

Proof-as-auth inserts one step into the path ahead of the moment a work using a voice is generated or published: it fixes what the use was checked against. It is not that a machine judges whether the voices sound alike. It puts “whose consent, and which grant, was this use authorized against” into a form in which the person on the receiving end can establish it — before the work is distributed or monetised, and without querying the issuer.

Lemma’s design against this gap:

• The consent it was checked against, fixed. Tie the generated work to the grant actually relied on — whose voice, for what scope, until when — and keep which grant it rests on in a form that can be checked before publication.
• A record of the check. Keep when that binding was made, under whose issuance, and that it has not been altered since — tamper-evident, in a form that cannot be overturned later. The claim that permission was obtained stops being merely a claim.
• Authorization before publication. Let distribution and monetisation proceed only where that check clears, so a work that carries no such proof can be held before the action.
• Only the result disclosed. Without handing over the whole of a person's consent record, make just the result — this use is covered by a grant — verifiable.

What this layer does not carry is worth stating as well.

• It does not judge whether two voices sound alike. Similarity is for a person to weigh, under the test the report sets out.
• It does not detect or remove generated works. Detection and takedown belong to the mechanisms already in place and to the exercise of rights.
• The gate on publishing sits with the platform's own process; this layer supplies the material for that decision, no more.

This is also where it differs from an operator’s own consent-management log. A log is something the party collecting the grants produces for itself; neither the rights holder nor the platform can check it independently.

Lemma does not replace legal protection. The damages, injunctions, and takedowns the report sets out are complementary to this layer, not alternatives to it. The first corrects harm after it has spread; the second closes one point before the work reaches the world.

Sources

• Ministry of Justice, Japan (primary, study group final report): “Study Group on Civil Liability for the Unauthorized Use of Likeness, Voice and the Like — Final Report: Interpretive Guidance on Infringement of Publicity Rights by Generative AI” (August 2026) — https://www.moj.go.jp/content/001468286.pdf
• Ministry of Justice, Japan (primary, study group proceedings): sessions one (24 April 2026) through five (27 July 2026) — https://www.moj.go.jp/MINJI/minji07_00400.html
• Japan Actors Union (stakeholder body, official position): “The Union’s view: on the Ministry of Justice guidance concerning infringement of publicity rights by generative AI” (2026-07-28) — https://www.nippairen.com/about/post-moj-guideline-2026.html