Grok kept talking about ‘white genocide’ due to an ‘unauthorized modification’

On May 14 at the chair’s Chatbot GROK repeatedly responded to x users In case of no connection of their surveys, he cited the allegations about the “White Genocide” in South Africa. Now, a statement Placed on social network, Elon muskThe AI ​​has caused “a certain reaction to the political issue” in the X “Unauthorized Modification” in the X “Unauthorized Modification”. He did not say what happened to the personnel who were disseminating the naughty update. However, this reported the “internal policy and basic values” of the modification and conducted a comprehensive investigation into the incident.

Different users of the website were placed several institutions that GroK was included in the controversial claims white South African farmers Faced with racial discrimination and land shortages in their country. Questions? Well, someone in a twole asked how many times the HBO streaming service. In others, the user asked a baseball player’s salary history. Someone else asked for more information about the WWE match. CNBC He was able to repeat Chatbot’s answers with white genocide references. The news website said, “When asked if there was a special programmed for the Promotion of the White Genocide, it was not a groc and” the cause and evidence is to provide actual, useful and safe answers. “

Before you answer Xai, Open Chief Sam Altman Snarky sent the answer X. “I am sure that Xai will soon give a complete and transparent explanation,” he said, then imitated GroK’s answers by talking about the white genocide. Xai, from now on, this will broadcast the system’s certificate for the public to give an opinion on each change. The company also said that Khai employees will not change the instructions of the GROK, which will not change the instructions of the GROK. In this case, which can recently fix it, it was able to instill the current inspection process. In addition, the company said that a team that can watch the events related to the response of the events that are not caught by automated systems.

Like Techcrunch He noted that this is not the first time that Xai is accused of a controversial grock behavior against an unauthorized change. In February, ChatBot’s short censorship sources and how to musk and how it was President Donald Trump Invalid information spreads. Xai co-founder Igor Babuschkin said the moment he was a naughty worker pushed A modified modification of Grok’s wishes.

If you buy something through a link in this article, we can win the commission.