All the images you will see in this article have been created with Dall-e, an artificial intelligence platform. All the images in this article are flying around the internet right now without copyright and, even if you wanted to, you couldn’t make a profit from their creation. The reason? It is simple. They are not considered works (generated by a human, in this case), so the intellectual property law of any country does not support their protection. In other words: the machines are not yet worthy of this right and, therefore, of charging for it.
The humans who have taken part in its creation are, ultimately, simple pawns in a process that still has a lot to legislate. The simple fact of having put a phrase in a software and waiting for the result to come out does not make me the owner of anything. And although there are several countries that already accept Artificial Intelligence patents, they are not the keynote in a much broader debate.
Artificial intelligence and its millions of uses and options have opened a new door in the legal field. One that has stirred something that had been established for several centuries. In the field of images and their creators, the issue is complicated. How to compare the product created by an AI system that drinks from a visual database configured by algorithms with the great artists? Since the first intellectual property laws were born at the dawn of the French Revolution, back in 1789 –as a response to safeguarding enlightened thought and the nascent role of the human being as a creator of value–, the need to rethink the idea had rarely been seen.
The arrival of Dall-e, Dall-e 2, Midjourney, and a long list of variations of AI imaging software have brought this debate to the fore. One in which the role between creator and owner is diluted in data and players represented by the Internet and technology.
Now, and with no solution on the table –beyond the regulation on Artificial Intelligence that is expected from the European Union at some point in 2023–, a new terrain is opening up that many are already beginning to consider. And it is that when there is money involved, things always tend to take on much more importance.
My image created with Artificial Intelligence is not a work, nor will it ever be
What is a work? This question could be a first-year philosophy question, but in this case it appeals directly to how the intellectual property rights of an image are understood. And more specifically those created with Artificial Intelligence.
Beyond the beauty and quality of it, Any work that has been generated by a human being is considered a work. Or, failing that, there is a free and creative choice of a human through a platform or technological system. If you meet these requirements, you will be covered by any Intellectual Property Law.
Dall-e or Midjouney are, in short, neural networks that learn from images without limit. Losing the originality of the creator (human, in this case), they only respond to very specific requests. Whether they turn out better or worse depends, to a large extent, on the ingenuity of the request. But of nothing else.
In any case, this is the current state of the matter and that it has to face a regulation that, at least, makes things clear from now on. “We are talking about an essential element for the development of our economy (Artificial Intelligence), so at least we will have more or less defined initial rules with which to manage it,” he explains. Albert Agustinoy, Intellectual Property Specialist in Cuatrecasas. On this, explains Leandro Núñez, a partner at Audens and specialized in technology and intellectual property, points out that “the most likely and most reasonable thing is that this will change in the short or medium term, since it is not fair that those who develop these tools do not have no protection.”
The background: a monkey photographer and a designer in the United States
Many will remember that macaca that, in an oversight of a wildlife photographer, took a selfie. It is probably the most famous selfie in the short history of this term. The ape, who was logically playing with her recent discovery, managed to take a photo with a more than remarkable result. And what started as an anecdote from the networks – at the moment in which the photographer published the photo and the story behind it spread like wildfire – ended up being a legal problem with her own entry into Wikipedia.
With the clear objective of profiting from everything that had been born as a result of the selfie, the photographer wanted to collect the image rights of the document. The reality, and this was demonstrated by a court, is that the photo had been taken by an ape. The fact of not being human, and the null intervention of the photographer in this case, left a clear path: it was not a copyrighted work and therefore its rights were technically free.
“As it is not the product of the human intellect, it has no protection,” explains Núñez, who uses the example of the ape as a way of understanding a situation that now affects images by Artificial Intelligence. A very different question, she points out, would be whether a photographer programs a camera to take a picture under certain circumstances; in this case, he would be the owner of the result.
You don’t have to go that far either. Already in 2018, in a legal process that lasted a full year, programmer Stephen Thaler tried to register a work created by an AI system. Although Thaler wanted to put it under the ownership of the algorithm that had intervened, the response from the Intellectual Property office was the same as for the ape influencer: there was no human hand, so there was nothing to record.
Database system to manage Artificial Intelligence
As is logical, the fact that at this time there is no regulation that covers everything originated with Artificial Intelligence does not mean that it is the correct path. There are already many debates that are held about something that will have to happen in a short time. And, as Núñez points out, the only exception to current Intellectual Property may be the key: databases.
In the 1980s, and with the birth of the first digital databases, the question of how to manage their ownership arose. They were not works, since they fed on information from third parties in the form of lists, It wasn’t really human either., since there was a digital hand involved. As a split from what had been born in the French Revolution –and that no one imagined would go so far– the sui generis right of databases arose, which protects the effort to create and manufacture them. With only 10 years of protection (compared to the vital plus 70 years of original protection), it’s a way for manufacturers to protect themselves. In this group are all those apps or digital systems that were born at that time. Microsoft wasn’t going to give up the rights to World for the world.
It is precisely this split that opens a good path for some experts. That, to a large extent, they think that the property rights of Artificial Intelligence should go along these paths. Protecting manufacturers, but not much. The challenge, in any case, is to unify all of this, just as the Berne Convention does to make copyright global.
Regardless, you’ll probably never own the rights
Bearing in mind that my images produced with Artificial Intelligence are not considered works, and as the experts point out, it remains to be seen what paths end up tipping the scales of regulation, there is another question about the air who would be the final owner of the images? Probably, I – who made the request to a platform for a medium – will never be. It will be said platform –Dall-e, in this case– or the database that feeds its neural network.
And the question is quite clear. My effort, as a demander of a specific image, has been minimal. Again, it depends a lot on my ability to order something and have the result I want. But it is the database, and the algorithms that learn from it, that generate the computational effort.
“The logic is that the creator of the platform has to be protected more that has trained the AI because there is more effort than the one that puts the sentence”, explains Núñez, or failing that of those that feed the system. And he adds the example of works by commission, which is, to date, what that can be most similar to this circumstance: “The author is the owner of the rights because he has created the work no matter how much I have told him what to do”.