Almost without realising it, prompts have gone from being complete strangers to becoming part of our technological vocabulary. All of us can ask Chat-GPT questions, interact with it, and thus generate prompts. However, as the complexity of the task asked of the tool increases, or as greater specificity of the results is required, the prompts must also be more precise and sophisticated. In these cases, we turn to professionals –prompt or request engineers– or even prompt banks. This is indicative of the economic value that may be attributed to these tasks, which are increasingly in demand.
In view of the foregoing, the question arises as to whether the instructions given by the user to AI to generate the desired content, rather than the very content generated by AI, which in principle would not be protectable since it is not a human creation, can be protected through intellectual property.
Protecting a creation through intellectual property requires having a literary, artistic or scientific work that is original in the sense that it is the author’s own creation and an expression of their free and creative decisions. Bearing this in mind and considering the current legislation, the answer to this question, a priori, would be negative. It is difficult to support a statement indicating that there is originality in a simple command, unless the command itself is original in form, which does not seem to happen very often. It can even be reasoned that a prompt is closer to the description of an idea, to the listing of some ingredients that we seek to find in the final content, than to a work itself. And, moreover, knowing that intellectual property does not protect ideas, we would again arrive at a negative answer.
Nevertheless, there are cases, such as computer programs, in which these requirements have been interpreted in a broader sense. In the 1990s, when the value of software as a commercial asset was clear, the decision was made to include source code under the umbrella of copyright protection. To do so, a great conceptual leap had to be made, recognising that, exceptionally, works that were not characterised by their artistic or creative value, but rather by their usefulness or functionality could be protected in this way. This same reasoning could be applied to prompts.
Furthermore, there are assets such as databases that have been included in the catalogue of protectable assets on the basis of criteria other than originality, such as the economic or human effort made to create said content. In this vein, it could not be ruled out that a series of sophisticated prompts could receive some type of protection through a right related to copyright. A right that does not currently exist, but could be created in the future if a need for protection that must be met is found.
A third possibility exists where the value attributable to a sequence of instructions would be reflected not in a monopoly over the prompt itself, but in the granting of a copyright on the resulting content to the user. This would be the same as understanding that a user who demonstrates highly creative behaviour through their messages is in fact a creator using technological tools to carry out their work.
This solution has recently been applied by Chinese courts. The ruling in the case of “Li v. Liu” last November held that the user is the author of the image called “Spring Breeze Brings Tenderness”, which was generated by AI following their instructions. In this case, it was proven that the author introduced around 30 prompts and more than 120 negative prompts to model the result of the generated image.
Contrary to this reasoning, however, we have the United States Copyright Office (USCO) and its ruling in the case of “Zarya of the Dawn”, which recognises that prompts can influence the generated image, but do not directly and unequivocally lead to a certain result, which is indeed unpredictable. Moreover, the fact that the same “prompt” can give different results every time indicates the presence of a component of randomness in the tool’s operation, thus making it impossible to uphold the claim that the work was created by the user.
Technology transforms the world, and this constantly requires us to rethink standards. But that does not mean changing them –I do not agree with the axiom that it is necessary to destroy in order to rebuild, since there are rules and principles that have been in place for centuries– but it is essential to ask ourselves how we should apply them to new realities. Where we must maintain continuity and where we must innovate. The debate is open.
Violeta Arnaiz Medina
Director of Technological Intellectual Property at PONS IP
