Qualification as a work created with the help of artificial intelligence and ownership of rights
The work is the result of a creative process; it is its individual and unique character that makes it a fully-fledged work protected by copyright, regardless of the means used by its author or its degree of originality, however modest.
The lengthy debated question in connection with the use of AI is the qualification of a work when it is created using an AI tool, and indirectly also the question of ownership of the underlying rights.
AI must be at the service of the creative, not the other way round. Traditionally, every work is born of a creative process on the part of its author, which then leads to the materialization of the work so that it can be protected. This classic distinction between idea (corpus mysticum) and form (corpus mechanicum) has made it possible to distinguish between the idea, which cannot be protected, and the formal creation, which can be protected by copyright. While the classical works of fine art are not called into question, an evolution has already taken place with the advent of photography. Initially disparaged, photography has imposed an indirect relationship with nature by recording a subject in a fixed image, using a range of techniques, processes and materials. And it is precisely the photographer’s choice (subject, framing, light, grain, exposure time, print, etc.) that makes it a unique and original work. The notion of a work of art has already been broadened on this occasion.
The notion of a work must therefore be constantly revisited and reinterpreted, particularly with the increasing digitization of society: a photograph recorded solely in digital format is protected as such in the same way as a paper print, even if it can be duplicated ad infinitum. According to Swiss law, the legislator has not made use of the possibility offered by Art. 2 para. 2 of the Berne Convention to require, as a condition of protection, that works be fixed on a material support. A work can thus exist even without being fixed on such a support (art. 29 al. 1 LDA), but the work must nevertheless be perceptible to the senses.
This interpretation of the notion of “work” has been exacerbated by contemporary art: because of the requirement of a formal creation, traditional jurisprudence and doctrine theoretically exclude concepts and ideas from the scope of protection. Nevertheless, contemporary works, and particularly conceptual ones, are characterized by the dimension given to the idea, which is at the heart of the creation. Indeed, it is the idea that gives the work its originality, the imprint of the author’s personality, the choices he or she has deliberately made. The protection of the work inevitably lies in the concept. If “the idea is self-sufficient”, it can be inscribed in a pre-existing form, which then becomes the random support for the idea; and it is sufficient that it has been “chosen” by the artist, not “created”. From then on, creation can indeed crystallize in the idea. However, if the work is dematerialized, the idea is even more present, since it is the only representation of the work[1] .
As a result, the difference between idea and form would be one of degree, rather than nature[2] , which would be measured by the yardstick of originality. A concrete work is one that is materialized into an entity perceptible to the senses; the form of the work then resides in the realization of an idea, its implementation by means of such and such elements, tangible and intangible[3] . Copyright thus protects the content of the creation in the concrete form given to it[4] .
If we retain these principles of AI-assisted creation, we can conclure, for example, that the choice of a subject – which can be expressed by a prompt (the “directives”) given to the machine by artificial intelligence – already constitutes a creative process; indeed, the author has an idea in his mind that he wishes to realize, and his instrument is then neither a paintbrush, nor his mouse, but the AI algorithm. In this way, the choice of a result is simply the continuum of the creative process. The recording or downloading of the result then constitutes the “tangible” or perceptible result required by law.
This is precisely why photocopies, scans, prints from negatives, prints of works saved in digital form, photos from surveillance cameras, radar, camera traps or webcams, as well as photos of two-dimensional objects (writings, plans, drawings) do not constitute works, as there is no room for manoeuvre, especially on the conceptual level.
The work is thus created. But to whom does it really belong?
Generally speaking, the digital world and the Internet are governed by private entities and regulated by general terms and conditions. Intellectual property rights arising from the use of AI tools are generally given very careful consideration when drafting terms of use for AI tools.
The AI behemoths have generally drawn up general terms and conditions that are non-negotiable by a (small) agency, limiting their own liability to the maximum, and in effect giving no guarantee to the user that they will not be accused of infringing third-party rights.
OpenAI’s terms and conditions applicable to all of its services state, for example, that[5] : You may provide input to the Services (“Input”), and receive output generated and returned by the Services based on the Input (“Output”). Input and Output are collectively “Content.” As between the parties and to the extent permitted by applicable law, you own all Input. Subject to your compliance with these Terms, OpenAI hereby assigns to you all its right, title and interest in and to Output. This means you can use Content for any purpose, including commercial purposes such as sale or publication, if you comply with these Terms. OpenAI may use Content to provide and maintain the Services, comply with applicable law, and enforce our policies. You are responsible for Content, including for ensuring that it does not violate any applicable law or these Terms.
This means that the user is made responsible for ensuring that the “content”, which also includes the “output”, does not infringe the rights of third parties. But OpenAi also exempts itself from any responsibility for the possible infringement of third-party rights to the output :
THE SERVICES ARE PROVIDED “AS IS.” EXCEPT TO THE EXTENT PROHIBITED BY LAW, WE AND OUR AFFILIATES AND LICENSORS MAKE NO WARRANTIES (EXPRESS, IMPLIED, STATUTORY OR OTHERWISE) WITH RESPECT TO THE SERVICES, AND DISCLAIM ALL WARRANTIES INCLUDING BUT NOT LIMITED TO […] NON-INFRINGEMENT […].
For the source code generator (Codex / Code Generation), the tool publisher has included a full disclaimer regarding the usability or otherwise of the output: Output generated by code generation features of our Services, including OpenAI Codexmay be subject to third party licenses, including, without limitation, open source licenses.
In plain English, this means that the publishers of AI tools do not guarantee that the results obtained do not infringe the rights of third parties, and are therefore subject to licensing conditions; furthermore, the publisher reserves the right to use the customer’s input to improve the tool (in other words, to train its algorithm – which is certainly fair game, but implies that the customer is obliged to assign a certain license on the content it creates, if this content is original). In some cases, pay-per-use versions of certain tools allow you to opt-out in this respect too.
**
This makes it difficult for creative agencies to guarantee that the use of AI can be carried out without risk, with regard to the use of the results obtained and their integration into the products and services produced for the agency’s clients. As indicated in a previous blog post, transparent information and customer agreement to the use of s artificial intelligence tools (via a mention in the general terms and conditions) will therefore be particularly indicated.
There is also no substitute for vigilance and common sense, to avoid the graphic designer being held responsible for the infringement of works he or she has never seen. It is therefore advisable to check the results obtained from the use of artificial intelligence tools, but also not to use them as they are, but only as a source of inspiration, modifying them sufficiently to add an identifiable personal note that is sufficiently different from the proposed results.
P&TS will be happy to discuss the terms of use with you in greater detail, and to recommend limitations to be included in your contractual terms and conditions.
[1] Mathilde MARCHAL , L’APPREHENSION DES NOTIONS DE FORME ET D’ORIGINALITE DU DROIT D’AUTEUR PAR L’ART CONTEMPORAIN , mémoire de master, UNIVERSITE D’AIX-MARSEILLE 2018-2019, P.41 et réf. cit. aux notes 200-202 ;
[2] CHERPILLOD I., L’objet du droit d’auteur, étude critique de la distinction fond/ forme, ed. Cedidac, 1985, n° 110, p. 53
[3] CHERPILLOD I., L’objet du droit d’auteur, étude critique de la distinction fond/ forme, op. cit. p. 74.
[4] Barrelet/ Egloff, Le nouveau droit d’auteur, Commentaire de la loi fédérale sur le droit d’auteur et les droits voisins, 4ème éd. Berne, 2021, N 19, p. 17 and ref citées
[5] https://openai.com/policies/terms-of-use