Table of contents
We have already discussed how generative AI works. Regarding the legal implications of the use of AI, in principle, we find two main problems: on the one hand, on the pre-existing content; and on the other hand, on the generated content.
On pre-existing content
It is clear that the use of protected works to feed a database of an AI solution, which generates new works, constitutes an infringement of the copyright of the owners of the pre-existing work, provided that it is used without their authorization.
Without prejudice to the possible infringement of the moral rights of the author (right of paternity, right of integrity of the work, etc.), this use also involves the infringement of the exploitation rights of the work, in particular the reproduction rights. Reproduction (art. 18 SCA) is understood as the exclusive right (authorizing or not) of the author for the fixation of the work in any medium and support that allows its communication or the obtaining of copies.
In principle, there would be no infringement of the right of transformation of art. 21 SCA, since this right includes that, from the result of the acts of modification of a pre-existing work, a different work is derived. The exercise of the right of transformation requires a creative effort by the subject and the originality of the result, which would turn the resulting work into a new work regulated in art. 11 SCA as a derivative work.
However, there will be no such infringement when the AI does not feed on pre-existing works and is limited to training the data models and executing algorithms with elements that are not protected by intellectual property law.
Here we can mention the case The New York Times vs. Open AI and Microsoft. The newspaper argues that millions of articles were used for information ingestion by AI. OpenAI collects this information in order to feed its data models and thus provide solutions to the user's prompts in the most suitable way possible and generate the requested content.
In short, the use of protected works to feed the database of a generative AI solution constitutes a copyright infringement of the owners of the pre-existing work being used as long as this act is performed without their authorization. Therefore, it will be a requirement that AI solutions obtain the corresponding authorization or rights license from the rights holders; as has recently happened with the works of Edith Piaf, whose voice will be recreated by artificial intelligence for the creation of a biographical film.
Legal implications of content generated by artificial intelligence
In Spanish law, copyright protection is only granted to those works that have been created by a natural person, i.e. by the work of human intelligence (art. 5 IPL).
Notwithstanding the above, the Law introduces a caveat with respect to the so-called “collective works” (art. 8 SCA), a machine cannot be considered an author, just as it is not possible to attribute the status of author when creations are the result of the actions of natural agents or animals, since the event giving rise to copyright is the creative work and not the discovery.
It is worth recalling, in a similar sense, the famous case of the monkey selfie. This is a photograph accidentally taken by a monkey that was the subject of an application for registration with UPSTO (the U.S. Copyright Office), which ruled that works created by animals do not qualify for copyright protection. This same fact can be extrapolated to machine-generated works.
So how do we overcome this legal regime?
The issue of works carried out autonomously by machines is a subject that, for years, has been controversial in the legal doctrine. However, currently the legal framework allows to assume all those questions that may arise on these works.
To do so, we must take into account the premise that, in order for a work to be considered the object of copyright, it must present the concurrence of two cumulative elements:
1. The existence of an original object that constitutes an intellectual creation of its author -understood as a natural person who performs an intellectual creative activity that is relevant and not insignificant with respect to the result-;
2. And the existence of an object identifiable with sufficient precision and objectivity.
Without prejudice to the current regulatory framework, the doctrine raises a series of foreseeable scenarios that, in the future, may come to govern the irruption of this technology:
1. Conservative option: deny intellectual property protection. This is the current position of the US and Spanish regulations, for example.
2. Disruptive solution: grant a legal personality of its own. This would be the so-called “electronic personality”.
3. Intermediate solution: grant a sui generis right for these creations. It would presumably imply a right in favor of the natural or legal persons who have coordinated the creation of the work and its disclosure as their own, (including the computer programmers who have developed the machine that in turn generates the work of art). This would require the amendment of the aforementioned art. 5 IPL.
4. Solomonic solution: ownership of the results obtained from the use of AI solutions is attributed to the authors or owners of the rights of the transformed works from which they originate in a co-authorship regime (collective work or collaborative work).
Additional assumptions
What happens if there is creative input by a natural person on AI-generated results?
The situation is different when, for example, a text generated through a generative AI solution is the subject of a creative contribution by the user, i.e. the user adapts or alters the original text in any way, creating a new work (derivative work).
This new creation will be considered a work, since it is the result of the intellectual contribution of a natural person on the result generated by the AI and will comply with the above-mentioned requirement of originality.
However, when the changes introduced in the text generated, for example, by ChatGPT are not sufficiently relevant (and lack, therefore, creative height and originality) we cannot consider that the result is protectable by copyright and, consequently, acquire exclusive rights of use over it.
Consequently, we can understand that a novel written as a result of the texts generated by ChatGPT can become copyrightable provided that the user performs on these a creative work and / or intellectual contribution.
Originality is in the prompt
It is certainly possible to find human creative intervention in the information transmitted to the artificial intelligence solution through the prompt. This process is similar to that of an employer providing specific guidelines to an employee for the realization of a collective work or when a principal gives instructions to a contractor for a contract work.
In this scenario the human creative intervention is at the conception stage of the work, which means that the more detailed and precise the instructions provided to the AI solution on the formal elements of the result, the less room is left for the creative execution of the machine, allowing us to observe that the work of the machine is essentially mechanical and lacks any originality.
Our lawyers, experts in intellectual property and the use of artificial intelligence, can help you with any questions on the subject. Contact them here.
Lawyer in the area of intellectual and industrial property.
Add new comment