Cloned AI voice: 4,000 euros for Bruce Willis' dubbing artist

Two videos, one cloned voice

Manfred Lehmann is the German voice of Bruce Willis. A YouTuber published two videos featuring an AI-generated voice that was virtually indistinguishable from Lehmann's. He had not asked him. In a judgment of 20 August 2025 (case no. 2 O 202/24), the Regional Court of Berlin II ordered the YouTuber to pay 4,000 euros. The court's key finding: an imitation created by AI also infringes the right to one's own voice. The requirement is that the audience attributes the voice to the real speaker.

The videos were satirical criticism of the German federal government's policies. At the end, the YouTuber advertised his online shop. There was no indication that the voice had been artificially generated. In the comments, viewers recognised the voice, and some named Lehmann.

Lehmann sent the YouTuber a cease-and-desist letter. He signed a declaration to cease and desist, but refused to pay. So Lehmann sued for 2,000 euros per video plus reimbursement of his legal costs.

The right to one's own voice

The general right of personality protects a person's name and image, and their voice as well. The right to one's own voice means that everyone decides for themselves whether and for what purpose their voice is used. The court acknowledged that the videos did not feature Lehmann's actual voice, but an AI imitation. That made no difference to the outcome. Legally, it is irrelevant whether a human or a piece of software imitates a voice.

The decisive factor was recognisability. Because of the similarity, a significant part of the audience assumed that Lehmann himself had spoken the text. The court called this "confusion of attribution": viewers could believe that Lehmann stood behind the statements or had consented to the use.

On top of that came a risk to his reputation. The court placed the YouTuber politically as "apparently rather right-wing". Without any indication that this was an AI voice, that association could damage Lehmann's standing.

Satire does not help the YouTuber

The YouTuber argued in his defence that he had merely been looking for an "authentic voice with a heroic sound" and had taken the voice the software suggested to him. The court rejected this. A human voice impersonator with a Bruce Willis sound would have exonerated him just as little.

Freedom of expression and artistic freedom did not help him either. The court weighed both fundamental rights against Lehmann's right to his voice. The satirical criticism of the federal government would have worked just as well with any other voice, and it was not directed against Lehmann himself. His voice served as a draw to generate reach and boost sales in the linked online shop. That is commercial use, and commercial use requires permission.

How the court arrived at 4,000 euros

Lehmann was awarded a notional licence fee, that is, the fee the YouTuber would have had to pay for permitted use. Whether Lehmann lost money because of the videos was irrelevant. Anyone who appropriates the economic value of a prominent voice without permission is unjustly enriched under the law and must hand that value over. As a benchmark, the court referred to a decision of the Federal Court of Justice (BGH) on unauthorised advertising using a picture of Günther Jauch, for which 20,000 euros became payable.

On the amount, the court heard a witness who has been booking Lehmann as a voice artist for many years. According to him, Lehmann is the most booked advertising voice in Germany, and his fees for advertising start at around 1,800 euros. In view of the channel's reach and the indefinite publication, the court considered 2,000 euros per video appropriate, so 4,000 euros in total. On top of that come the legal costs of the cease-and-desist letter. When it was handed down, the judgment was not yet final; the YouTuber could still appeal.

Consequences for users of AI voices

For creators, podcasters and freelancers who use AI voices, the judgment holds three lessons:

• The sound is what counts. How the voice comes across to the audience is the decisive question. If it sounds like a well-known speaker, it makes no legal difference whether a human or a piece of software imitated it.
• The software's suggestion offers no protection. Anyone who takes a ready-made voice from a generator bears the risk that it was modelled on a real person.
• Commercial use carries weight. As soon as a video is meant to generate reach, contains advertising or links to a shop, the court's reasoning applies.

The lack of any indication that this was an AI voice did the YouTuber further harm, because viewers were bound to assume that Lehmann stood behind the statements. The court did not have to decide whether labelling alone would have made the use permissible. In any event, the EU AI Act requires AI-generated voices and deepfakes to be labelled. The safest course is to use your own voice, licence a voice with clear usage rights, or ask the speaker for permission.

What those affected can do

For voice artists, podcasters and anyone else whose voice someone clones without permission, the case shows the way. The first step is a cease-and-desist letter demanding that the use stop. The second is a claim for a notional licence fee, calculated on the basis of your own usual fees. Evidence of recognisability is important, for example comments in which viewers identify the speaker.

In Lehmann's case, his fame helped. After the judgment, his own lawyer pointed out that with less well-known speakers, attribution by the audience is harder to prove. Lehmann said his aim had been a landmark ruling for the whole industry, and at first instance that is what he got.