The user asked, the AI performed: who bears the responsibility?

In the USA, a lawsuit was filed against the company xAI, owned by Elon Musk, which was joined by a young woman who claims that her stepfather used the artificial intelligence application to create pornographic images based on photos he took when she was a child.

This is not Musk's first entanglement in a lawsuit of this kind: a few months ago, Musk and "Grok" were accused of allowing users on Twitter (X) to use the software to create sexual images based on ordinary photos of children.

At least in the case of the young woman from Wyoming, a nurse by profession, the publication had devastating effects beyond the personal and family embarrassment. There seems to be no dispute about the extent of the damage, and certainly the emotional distress suffered by the victims. The question is whether Musk's companies or technology companies in general should have responsibility for the harm experienced by those who became pornographic objects against their will.

This question is fascinating, not only because it will have far-reaching consequences, on the scale of hundreds of billions of dollars for the artificial intelligence industry, but because it brings to the table philosophical issues from the field of morality and law that we have not had to deal with in the past.

Let's start with what is undisputed: the degree of responsibility of those who forge and distribute the pornographic content. They are certainly guilty, but should the technology that served them also bear responsibility?

Apparently, the immediate and simple answer is "no". When a person is stabbed to death, the stabber is sued, not the knife that allowed the crime to be committed. The same is true for a gun or any hardware that allowed physical harm to another person. But what happens when what is supposedly just a tool for committing a crime includes not only the execution but also the planning? And even then, where exactly is the line drawn that imposes part of the blame on the software owners?

Sports car

For the purpose of the discussion, let's go for a moment to an intermediate stage, with a tool whose features allow for a bit more than a knife stab or a gunshot. We are talking about a vehicle. Our cars can accelerate to a speed of nearly 200 km/h. It is clear that at such a speed almost any accident will be fatal, so why not impose part of the responsibility on the car manufacturers? Not only do they not produce tools that allow the level of risk to exceed the level of safety — for example, a chassis that would withstand even a head-on collision at high speed, they even cultivate with the help of advertising and public relations a system that sometimes encourages careless driving.

For what are those performances expected from the sports car and the driver, if not an invitation to test the limits of the capabilities of both? And yet, although on the face of it there might be a contribution to the conditions that allow for road accidents, car manufacturers are not sued as causing road accidents, but only in the case of a proven technical failure, for which the manufacturer is responsible.

So it is impossible to sue a manufacturer of knives, firearms, and vehicles, because otherwise, one could sue a manufacturer of almost anything: a meat tenderizer, a baseball bat, a hair dryer thrown into a bathtub, and all those less common means of killing that we see in Hollywood movies.

Accomplice to the crime

Now let's move on to those who can be sued in such cases — and we mean not only the mastermind of the crime, but also the one who carried it out. It is clear that one cannot sue a weapon manufacturer, but one can sue the one who pulled the trigger, even if it is a murder ordered by someone else. And that is good, because otherwise contract killers would be exempt from guilt and punishment just because they committed a murder on behalf of a third party.

That is, in such cases, both the person who ordered the crime and the perpetrator are put on trial: since the former initiated the offense, but the perpetrator planned the way and the time, or at least pulled the trigger — and thus became a full accomplice to the crime, without whom it would not have been committed.

And this is exactly the argument of those who are suing the artificial intelligence applications that allow the creation of pornography from ordinary photos. According to them, this is not a knife, a gun, or even a car, but full accomplices to the crime, like contract killers.

For the benefit of those who are not familiar with the secrets of AI and wonder what it has to do with Photoshop software, for example, since it too can be used to forge pornography, well, the difference is not only in the appearance, but in the degree of complicity in the crime.

In image processing software (like Photoshop), the software operator must perform the processing with his own hands and by direct instruction to the software with a mouse click. In contrast, in AI, the software is the one that creates the final image, not by an explicit action of the user but in an apparently independent way, based on his intentions as it interprets them. I can take any photo that exists on the network — and in the era of social platforms, almost all of us have photos that we uploaded from a vacation, a restaurant, work, and so on — and ask the software to give it a pornographic interpretation.

This is exactly what the plaintiffs are aiming at, claiming that the application, in this case, is not just a tool but an active partner in the process, from the planning stage to the execution stage. Are they right? Regardless of the verdict, which will obviously take many more years, this is a fascinating philosophical question, since artificial intelligence, unlike the human brain, has no malicious intent. While it is producing a pornographic image from the victim's photo, it has no ability to understand that it is a partner in a wrongful act.

This is only the beginning of the discussion, which raises philosophical issues that did not exist before the AI era, meaning judges will have no precedents to rely on. Admit that it is quite amazing: a reality in which not only the damages caused to the soul or reputation of the victims are placed on the scales of justice, but also the consequences of hundreds of billions of dollars for one of the most profitable technology sectors.

I do not pretend to decide, but only to state two facts: the first is that artificial intelligence can be a double-edged sword, but it is mandatory to remember that it is a great auxiliary tool that simplifies and makes excellent technology accessible to the general population.

The second fact is that the lawsuits filed against the owners of artificial intelligence tools are only the first swallows of the kind that will occupy the courts in the coming years. Dealing with them will require an approach that will allow changing the water in the bathtub, without throwing out the baby with the bathwater.