Toward a Right Not to Have Your Voice Used — Japan's Copyright Act Protects the Performance, Not the Voice

Copyright ActStatute exists. Voice is outside it

What it protects: A performance, and its recording and transmission

If a voice is imitated without recording any performance, neither the recording right nor the making-available right applies

Publicity rightNo statute. Outlines set by case law

What it protects: The exclusive use of the customer-drawing power of a name or likeness

No statute. It is judge-made, recognised as part of a right derived from personality; whether voice is included is not written down

Unfair Competition Prevention ActStill listed as homework

What it protects: Indications that cause confusion, among others

The 2024 IP strategy left the relationship with voice as homework for five ministries and agencies

On 27 July 2026 an expert panel at Japan's justice ministry broadly agreed a draft report stating that the publicity right should cover voice as it covers likeness. The sequence is to agree the outline before writing any statute. The people envisaged are public figures.

On 27 July 2026, an expert panel at Japan's Ministry of Justice broadly agreed a draft report. It states that the publicity right, under which a public figure's name or photograph may not be used without permission, should be recognised for voice as it is for likeness. The spread of generative AI lies behind it.

At a glance this looks like one more right being added. What it actually does is shift what existing law cannot catch onto a right that has no statute. Two gaps overlap.

The first is in the Copyright Act. The Act grants performers rights: to record or film a performance, and to make it available for transmission. What is protected there is the performance, not the person's voice. A paper the Agency for Cultural Affairs put to a working team sets this out squarely as an open question. Where no performance has been recorded, but a voice closely resembling a particular voice actor is generated by AI and made to sing an existing song in an audio file or video, do any rights under the Act reach it?

The second gap is in the publicity right itself. It is written into no statute at all.

The outline of the publicity right was drawn by a Supreme Court judgment. In a case brought by singers whose photographs a weekly magazine had published without permission, the First Petty Bench put it this way. A person's name and likeness symbolise their personality, so the individual holds, as something derived from the right of personality, a right not to have them used arbitrarily. A likeness may carry the power to draw customers and promote sales, and because the exclusive right to that drawing power rests on the commercial value of the likeness itself, it forms part of that right derived from personality.

The judgment also set out when infringement occurs. Where the likeness is used as a product to be appreciated in its own right. Where it is attached to goods in order to differentiate them. Where it is used as advertising. In such cases, where the purpose can be said to be solely the exploitation of that drawing power, the use becomes unlawful as a tort.

What may fall inside the phrase "name, likeness and the like" is not spelled out in the judgment. Whether voice can be placed there is what the panel is now discussing. This is moving as a question about the reach of judge-made doctrine, not as an amendment to any statute.

There is another route. Japan's Intellectual Property Strategic Programme 2024[1] stated that the relationship with the Unfair Competition Prevention Act would be clarified, and revisions considered if needed, for the use and generation by AI of the likeness and voice of actors and voice actors. Listed as responsible were the Ministry of Economy, Trade and Industry, the Agency for Cultural Affairs, the Patent Office, the Ministry of Justice and the Consumer Affairs Agency. Which is to say that two years ago it was not settled which law would take this.

Technology moves first and the law draws the outline afterwards. That order is not unusual in itself. The problem is that there are two ways to draw it, and the one in use here is the second.

The first is to write a statute. The legislature fixes the conditions and the consequences, and anyone reading it draws the same line. The second is to draw through case law. Somebody sues, a court rules, and the accumulation becomes a line. The publicity right grew the second way. Its outline therefore follows the shape of the cases that were brought.

That difference bites when you ask who is able to sue. Litigation costs money and time. Harm from having a voice imitated by AI can befall a well-known voice actor, an unknown streamer, or an ordinary person. But only those who can retain counsel and fight to the Supreme Court can make case law. That the panel envisages public figures is not because the need for protection is concentrated among them. It is because the procedure for drawing the outline can only begin with them.

The customer-drawing requirement pulls the same way. What the judgment placed at the base of protection is the power of a person's name or appearance to sell goods. A right stands for those whose voice sells. But commercial value is not the only reason having your voice taken matters. You are made to have said what you did not say. It is used in fraud. Your family believes it is you. None of these can be measured by drawing power. Harms that fit the yardstick are remedied first; harms that do not fit wait.

So what follows? Two orderings are available. Continue along the present route of extending the doctrine to voice, while separately building a vessel for harm to people who are not famous. Or write a statute, under unfair competition law or new legislation, that takes confusion or unauthorised use itself as the condition. The first is fast, but its outline depends on which cases happen to be brought. The second is slow, but the same line is drawn for everyone. What the IP strategy left as homework for five ministries is, in substance, this choice.

In needing to count separately what has been made available and what has actually arrived, this overlaps with policy exclusion and non-take-up（このサイトの記事）. On dividing responsibility around AI systems, risk management for AI agents（このサイトの記事） is also relevant. A voice is something a person has used since birth, requiring no registration and no renewal. Precisely because none was required, it was never written into any law. The question is whose case starts the procedure for protecting what was never written down.

Further Reading

• AIと社会と法 (AI, Society and Law)（外部サイト、新しいタブで開きます） (edited by George Shishido, Takehiro Ohya, Soichiro Kozuka and others, Yuhikaku, August 2020). A collection that treats the legal questions raised by AI across fields rather than within any single one. It gives a map for thinking about the territory, discussed above, that no statute catches head-on.

References

On the Relationship Between Copyright and the Generation and Use by AI of Voices Imitating Voice Actors (Material 2) — Copyright Division, Agency for Cultural Affairs (2024). Working Team, Legal Systems Subcommittee, Copyright Subdivision, Council for Cultural Affairs

Views on AI and Copyright (Summary) — Copyright Division, Agency for Cultural Affairs (2024). Legal Systems Subcommittee, Copyright Subdivision, Council for Cultural Affairs

On AI and Copyright — Agency for Cultural Affairs (2026). Agency for Cultural Affairs

A Right Not to Have Your Voice Used Without Permission: Justice Ministry Panel Drafts Report — NHK (2026). NHK NEWS WEB