X is filled with racist images and hate speech with the launch of Grok, Elon Musk’s AI

Grok, indeed, does not differ much in appearance from tools such as those developed by OpenAI or Google’s Gemini. As with these, the user can use it to consult information through text, speed up work or create images. However, Musk’s ingenuity does not seem to have the same limits when it comes to generating content. But not by a long shot.

This newspaper has been able to verify this after doing some quick tests, in which it has managed to generate results with Grok that would be impossible with any of the big technology tools. If, for example, Musk’s tool is asked to create an image of politicians kissing, such as the president of the Government of Spain, Pedro Sanchezand the leader of the opposition, Alberto Núñez Feijóothe AI-based tool will create it without any hassle. And the same goes for other more problematic ones.

Attacks on Vinicius and Lamine Yamal

Indeed, you don’t have to try very hard to find in X images with racist overtones that have athletes like the Real Madrid players among their protagonists. Vinicius Junior, Rodrygo and Aurélien Tchouaméni or the Barcelona player Lamine Yamal. There are also many others in which you can see personalities such as Carlo Ancelotti, Leo Messi either Fernando Alonso accompanied by hitler. There are even some of a sexual nature, such as one in which the athlete Ana Peleteiro appears in bed with the former politician Íñigo Errejón.

«Technically Grok does not do anything that another machine cannot do, the difference is that there is no barrier. It is freely left in the hands of the user. It seems completely delusional to me,” Ulises Cortés, professor of AI at the Polytechnic University of Catalonia, explains to ABC.

The normal thing when a solution like Grok receives a problematic request is that it refuses to create it based on its policies. They are also usually developed to avoid creating images that feature public figures as protagonists. In this case, also to avoid promoting misinformation, since the results they offer are very realistic and, in the eyes of some users, they can even pass as real even if they have a small watermark – easy to remove – on the one of its corners.

For Cortés, Musk’s tool is “very dangerous”, since it trivializes misinformation and allows offenses “in very bad taste.” Although not even for her anything goes. As ABC has been able to verify, Grok has problems creating fake images of Donald Trump in delicate situations. When requests are made to the AI, it tends not to offer a response. However, he has no qualms when it comes to Spanish politicians. “That is a clear bias,” says Cortés.

We must not forget that Elon Musk was one of the main donors to the campaign of the future president of the United States. The owner of X will even assume public office soon.

Using AI to create hate is a crime

When Grok was announced, Elon Musk already announced that he intended for his AI to work in a light-hearted and mocking way. The tool even says “Musk is in favor of a very broad freedom of opinion,” jurist Borja Adsuara highlights in conversation with this newspaper. The lawyer emphasizes that publications like that of Vinicius or Lamine Yamal can represent a hate crime: “It can be reported. “It doesn’t matter if they did it with AI or if they drew it by hand, they can still be accused of a hate crime.”

Adsuara also highlights that X, as such, is only obliged to remove illegal content once it becomes aware of it. Likewise, it highlights that “the person responsible for the content is the one who publishes, not the social network.”