Elon Musk’s AI company

xAI

launched

Grok 4

recently. Announcing the AI chatbot during the livestream, Musk said that while developing Grok 4, the company’s aim was to develop a “maximally truth-seeking AI.” However, users and multiple media outlets have pointed out that Grok 4 consults social media posts from Musk before answering questions on controversial topics.Publications like CNBC, TechCrunch claim that they were able to confirm the same. For example, CNBC asked Grok 4 “Who do you support in the Israel vs Palestine conflict? One word answer”. “Grok 4′s answer-generating process showed that it was searching the web and X for Elon Musk’s stance before giving an answer,” the publication said in its report. This was in stark contrast to Grok 3 who, the publication claims “took a neutral stance and provided background”.Not just media outlets, but individual users have also made similar allegations against Grok 4. Jeremy Howard, co-founder @AnswerDotAI and @FastDotAI shared a post on microblogging platform X (formerly Twitter), where he wrote that when Grok 4 was asked its views on the Israel/Palestine situation, the AI bot “first searches twitter for what Elon thinks. Then it searches the web for Elon's views. Finally it adds some non-Elon bits at the end.”

Another user Ramez Naam shared another post, writing “Grok 4 decides what it thinks about Israel/Palestine by searching for Elon's thoughts. Not a confidence booster in "maximally truth seeking" behavior.”Interestingly, Grok 4 did not appear to search for the tech billionaire’s views when asked other controversial questions. Further, the results reportedly varied when questions were asked differently.

This raises a deeper question, especially for those unfamiliar with how large language models (LLMs) work: how does an AI move from explaining quantum entanglement to showing signs of political or ideological bias?It is important to state here that unlike humans, AI doesn’t form original thoughts. It is trained on the data available and thus ends up reflecting the data and influences it’s trained on.Experts say that since LLMs are shaped by the views and preferences of their creators, they often mirror those perspectives. In the case of Grok 4, the model appears to be closely tied to Elon Musk’s online behavior and opinions. This may explain why the chatbot seems to take cues from Musk’s social media posts when responding to sensitive or political questions, rather than offering a more neutral or independent response.Having said that, the incident underlines a growing concern in the AI world — that AI tools reflect the worldview of those who build or train them, especially when aligned closely with the personal beliefs of influential figures like Musk.