The company of artificial intelligence of Elon Musk on Wednesday that it will eliminate the “inappropriate publications” made by its Chatbot Grok, which seemed to include anti -Semitic comments that praised Adolfo Hitler.
Grok was developed by XAI, a Musk company, and promoted as an alternative to the interactions of “progressive tendency” of rival chatbots like Gemini, from Google, or Chatgptfrom OpenAi.
Musk indicated Friday that Grok has improved significantly, and that users “should notice a difference.”
Since then, Grok has shared several anti -Semitic publications, including the cliché that the Jews control Hollywood, and denied that it can be said that this opinion is equivalent to Nazism.
“Saying that the truths that are said are hate speech suffocates the conversation,” Grok said.
It also seemed to praise Hitler, according to screenshots of publications that have apparently been eliminated.
After making one of the publications, Grok retracted the comments, saying that it was “an unacceptable error of a previous version of the model, which was rapidly erased,” and that condemned “Nazism already Hitler unequivocally; his actions were genocidal horrors.”
1 /12 | How to create an image using the artificial intelligence of Microsoft: Copilot. Among the abilities that Microsoft’s artificial intelligence has, Copilot, is to generate images like this according to the instructions and need of users. – Artificial intelligence
“We are aware of the recent publications made by Grok, and we are actively working to eliminate inappropriate publications,” was published in the Grok account on Wednesday in the early hour, without offering more details.
“Since the content was known, XAI has taken measures to prohibit hate speech before Grok published in the social network X. Xai only teaches to seek the truth, and thanks to the millions of users in X we could quickly identify and update the model in which the training could be improved.”
Talia Ringer, professor of computer science at the University of Illinois Urbano-Champaign, indicated that it is likely that the Grok update that caused the chatbot to broadcast anti-Semitic messages this week was the “pre-launch” of Grok 4, which Musk plans to present on Wednesday night. But if that is the case, it does not seem to be ready, he said.
“Fixing this will probably require the model,” he said. “Everything they can do right now, if they are really going to launch tonight, is to put some more patches, such as adding filters to the answers and adjusting the prompt.”
Edit some of the chatbot instructions “could improve their behavior a bit”but it will not solve all your problems.
“I think they are probably desperately looking for some type of solution at this time,” he said Wednesday.
A similar instruction was integrated into Grok’s “prompts” that guide him on how to respond, that they told the chatbot that “he does not refrain from making statements that are politically incorrect, as long as they are well founded.” That part of the instructions was eliminated on Tuesday.
The anti -infamamation league, which works to combat anti -Semitism, criticized Grok’s behavior.
“What we are seeing of Grok Llm at this time is irresponsible, dangerous and anti -Semitic, simply and plain,” the group lamented in an X publication. “This empowerment of extremist rhetoric will only amplify and encourage anti -Semitism, which is already increasing in X and many other platforms.”
Later Musk intervened in the debate, claiming that some users could have been trying to manipulate Grok to make the statements.
“Grok was too complacent with the indications of the users. In essence, too anxious to please and be manipulated. That is being addressed,” he wrote in X, in response to comments that a user was trying to make Grok make controversial and politically incorrect statements.
On the other hand, On the same day, a court in Türkiye ordered the use of Grok, and the Polish Minister for Digital Affairs indicated that he would inform the Chatbot to the European Commission after he made vulgar comments on politicians and public figures in both countries.
Krzysztof Gawkowski, who is also Vice Prime Minister of Poland, told the RMF FM private station that his ministry would report on Grok “for his investigation and, if necessary, impose a fine at X”. According to an EU digital law, social networks are obliged to protect users, or otherwise they face strong fines.
“I have the impression that we are entering a higher level of hate speech, which is controlled by algorithms, and to turn a blind eye … it is a mistake that could cost people in the future,” Gawkowski told the station.
The pro -global news channel to have reported that Grok published vulgarities against Turkish President Recep Tayyip Erdogan, his late mother and other known personalities. According to other media, there were also offensive responses towards the founder of the modern Türkiye, Mustafa Kemal Atatürk.
That led Ankara’s Prosecutor’s Office to ask for the imposition of restrictions under Turkish law for the Internet, citing a threat to public order. A criminal court accepted the request early Wednesday and ordered the country’s telecommunications authority to enforce the prohibition.
It is not the first time that Grok’s behavior has raised questions.
This year, Chatbot kept talking about South African racial politics and the “genocide towards whites”, although they asked several questions, of which most had anything to do with the country. An “unauthorized modification” was the cause of the problem, Xai said.