Use of a narrator’s voice without their consent in an AI-generated message and the infringement of personal rights

The use of content generated by artificial intelligence may have certain legal consequences. A company in the water and sewage sector found this out the hard way after broadcasting an advert featuring a message created using AI. It turned out that the voice of a well-known narrator’s voice had been used in an advert for domestic sewage treatment system without his knowledge or consent.

The owner of an online voice bank, whose database contained a sample of the narrator’s voice, brought an action to the Regional Court in Warsaw, 22nd Intellectual Property Division. The court proceedings are ongoing, and their outcome may have significant implications for case law in matters related to the use of artificial intelligence.

Polish law does not contain any explicit provisions concerning the protection of the voice; however, the basis for such protection may be found, on the one hand, in the provisions of the Act on Copyright and Related Rights and, on the other hand, in the provisions of the Civil Code concerning the protection of personal rights.

Article 81(1) of the Copyright Act prohibits the dissemination of person’s image without their consent. It is a matter of dispute whether this provision also covers the voice understood as part of the image, i.e. the perception of person through the sense of hearing.

There is less doubt as to  whether the voice qualifies as a personal right protected under Articles 23 and 24 of the Civil Code. This is because the list of personal rights is open-ended, and protection will apply regardless of whether the voice is classified as a separate personal right or as an element of the image.

In its judgment of 7 March 2023, case II CSKP 659/22, the Supreme Court held that the voice, like, for example, one’s image is a right encompassing a specific element of a person’s physical identity and, at the same time, is individualised due to its almost unique timbre and pitch. As such, it is subject to independent protection.

In the aforementioned ruling, the Supreme Court noted that, as a personal right, the voice is subject to independent protection and may be subject to interference, which may take the form of various actions that cannot, however, be categorised uniformly (a typical example of interference with the voice as an independent personal right is its alteration using digital tools in such a way that the timbre of the sound allows the listener to identify the voice belonging to another person, whilst the use of such a recorded sound results in the false attribution to that person of having uttered specific content – a phenomenon known as “deep fake”).

In earlier judgments, however, the Supreme Court expressed a different view, stating that, given that a breach of this right overlaps with a breach of other rights (reputation, dignity, privacy), there is no doubt that the voice should not be treated as a separate personal right. It should be protected within the framework of the right to one’s image, as it constitutes an element of that right not as a physical (visual) image of a person, but as an image perceived through hearing. The personal right to one’s voice should be treated as auditory image, provided, however, that it is recognisable to third parties. This means that the individual’s voice must be sufficiently characteristic and distinctive to allow for the identification of the specific person from whom it originates (see the judgment of the Supreme Court of 3 October 2007, case no II CSK 207/07).

In the case concerning the use of a narrator’s voice in an advertising message generated by artificial intelligence, it will therefore be necessary to determine, on the one hand, the issue of civil law protection of the voice and, on the other, the possibility of attributing liability to the entity using the AI-generated content. In this context, the position expressed by the Munich Regional Court in the case against Google may be relevant (for more details, see the article “Liability for AI-generated content”).