AI has a copyright problembut what about Elon Musk and Grok 2 has reached a new level. X has been filled with AI-generated images of celebrities. And although there are certain requests that cause errors, the truth is that it is possible to create images of celebrities in the most compromising situations. Musk explains that Aurora It is a model of “AI without censorship”, but what we have is a tool that raises numerous questions about its legality.
Grok breaks the rules. Unlike other companies such as OpenAI, X’s image generation allows you to have celebrity results directly. If companies like OpenAI are already facing lawsuits for using copyrighted contentElon Musk directly opts for the route of “asking for forgiveness before permission.”
Scarlett Johansson already warned. The use of images and content with image rights for AI training has been on the table for years. One of the famous actresses who has raised her voice the most against this use is Scarlett Johansson.
OpenAI had to withdraw a tone of voice which sounded too similar to the actress’s voice. The company argued that “it was not intended to be similar,” but they eliminated it to protect themselves legally. In the case of Grok, the similarity between the generated images and the image of the famous person is more than evident.
How have they achieved it? There are too many photos of celebrities on X. Since this summer, X decided to activate by default for all users to train the AI with their data. It was possible to disable itbut millions of users predictably have not done it or have not even known about it. The millions of images that have been published on Twitter have been used to train Grok.
While some generators are based on large image banks, the power of Grok is that it uses a database that only they have access to: X itself.
Is all this legal now? Elon Musk knows what he is doing. And what happens here is that in the United States there is no defined jurisprudence on to what extent is it legal use these images without rights to train AI.
Their defense is based on the fact that it is an acceptable use of “fair use”, a definition that does not exist in other countries, but does exist in the United States. According to they explain from X: “Certain uses of copyrighted materials do not require the permission of the copyright holder. The final decision as to whether a particular use of copyrighted material constitutes fair use rests with a court of law.” .
Musk has opted for a long-distance race. The entire industry is aware that this use of AI may not fall under legitimate use, but many years may pass until the Supreme Court determines this. Meanwhile, Grok delivers results that none of its competitors match.
The responsibility for misuse of these images lies with the user. As read in the xAI terms of service: “We take no responsibility for any entries. You will be solely responsible for your contributions.”
Terms that also reflect what types of uses of Grok are prohibited: “You must not use any results related to a person for any purpose that could have a legal or material impact on that person, such as making important decisions about credit, education, employment, housing, insurance, legal affairs, law enforcement, citizenship, health or other important aspects.
As Borja Adsuara points outa lawyer expert in digital law: “criminal images can be reported. It does not matter whether they were made with AI or drawn by hand. The person responsible for the content is the one who publishes them, not the AI or the social network.”
Is no celebrity going to denounce them? Grok’s new AI has been around for a few days and at the moment there is no evidence that any celebrity has filed a lawsuit against X. Yes, it would be the necessary step for the American justice system to determine to what extent it is fair use.
In Spain, seeing the large number of images, it would not be surprising if there was some demand for crimes against honor, privacy and self-image. Although, as we explained, this accusation would go against the specific users who have chosen to take advantage of this tool to do harm.
Regardless of whether there is damage, many celebrities are not inclined to let others take financial credit from their image. And in this management of image rights there could also be a lawsuit against X. One legal battle that seems very complex and is expected to only be available to high-level law firms.
Europe is already preparing its million-dollar fine. What happened with Grok is another reason for Europe to challenge a million-dollar fine against X. The European Commission has already warned Musk’s social network that has breached the Digital Services Law for the presence of “malicious actors with the blue check” and faces a penalty of up to 6% of its annual income. A sanction to which Musk replied that will face in court.
Under the Digital Services Law, X has the obligation to remove and provide mechanisms to flag illegal content. What with Grok they mean images of celebrities in situations that could compromise their honor.
xAI goes about his business. Another legal battle Grok has is with the training of his imaging model itself. In the United States this use could be covered by fair use, but in Europe the use of copyrighted images to train AI is prohibited in the new AI Law.
While more than 100 companiesincluding Google, Amazon, OpenAI or Microsoft, have agreed to comply with this regulation, xAI, the company behind Grokdoes not appear in the list. Musk will do his thing until the judges say otherwise. And that can take a long time.