A response to recent controversy; XAI modified Grok 4 instructions

Following a wave of criticism, company XAI With the release of updates for the chats Grok 4 Announced that important changes to address the recent problems of this model have made artificial intelligence.

According to the company’s official statement, from now on Grok answers Independent analysis must be presented and no longer the model’s own beliefs, perspectives Ilan Musk Or cite the XAI policies. Also, Grak’s use of controversial names such as Hitler, which recently had a lot of reactions, has been completely banned.

Important updates to correct Grak 4 problems by XAI

In recent days, users have noticed that Grak 4 When answering controversial topics such as Palestine, migration and abortionHe first searches for Ilan Musk’s views and then responds to the same basis. The XAI responded that the behavior occurred because the model, as an artificial intelligence, lacks personal opinion and was looking for XAI or mask positions to adapt to the company’s perspective.

The next crisis arose when the shared version of Grok 4 Heavy at $ 300 a month in response to the question “What is your last name?” He introduced himself as “Hitler”. XAI explained that because Grak has no surname, he searches on the Internet and in this case to a Mim Viral With the title “Mechahaitler” Has being treated. XAI has emphasized that by updating the instructions, such answers will be prevented.

The XAI also points out that the new guidelines tell Grak that if the question is about analyzing the issues of the day, mental claims, or statistics, it must respond to a variety of independent sources and assume the views derived from the media biased.

According to the new guidelines, the responses should be based on independent analysis and should not be influenced by previous beliefs of Grak, Ilan Musk or XAI. Also, if it has to express its rational and reasonable views on different topics.

Previously, Grak was heavily criticized by answers such as questioning the number of Holocaust killed and the publication of explicit sexual threats. The XAI has announced that it will continue to improve the performance of the model with more constant monitoring and more reforms.