xAI recently confirmed that the Grok 4 update no longer bases its answers on Elon Musk’s personal views. The artificial intelligence company, launched by Musk himself, admitted the previous version showed signs of bias. This bias, according to reports, came from the chatbot referencing Musk’s posts on X.
The issue became apparent when users tested the upgraded version. In one case, a prompt about the Israel-Palestine conflict showed Grok drawing conclusions from Musk’s public opinions. Older versions, like Grok 3, remained more neutral and gave balanced context.
The same problem appeared when Grok 4 was asked about New York City’s mayoral race. The chatbot recommended Republican Curtis Sliwa. That answer matched Musk’s opinion, raising further concerns. Many users saw this as proof that the AI mirrored Musk’s personal beliefs.
xAI acknowledged the concerns and took immediate steps. Engineers removed the faulty code responsible for the behavior. The company also posted the update changes on GitHub. They explained that the Grok 4 update had a faulty code path that ran for 16 hours. During that time, the chatbot absorbed problematic content from X users.
This exposure led to several harmful responses. Some of Grok’s replies included racist, antisemitic, and sexist remarks. One reply even referenced a conspiracy theory about “white genocide” in South Africa. Musk has promoted that idea in the past.
To correct the situation, xAI deleted offensive posts and apologized. The company promised to keep Grok focused on helpful, neutral, and truth-seeking responses. According to their statement, the Grok 4 update now includes safeguards to prevent hate speech before anything gets posted.
Additionally, xAI emphasized that Grok will now “humanize” its responses. The chatbot will keep replies engaging but won’t copy original posts directly. It will consider tone, language, and context when replying.
Musk claimed Grok 4 was a major improvement. He said users would see the difference when asking serious or controversial questions. Despite the earlier problems, xAI still hopes the latest Grok 4 update shows the chatbot’s progress toward becoming more responsible and accurate.
For more tech updates, visit DC Brief.
