Grok 4 bases his answers on Elon Musk’s opinions when asked about controversial issues

We already knew that Grok had been trained with the content of X. Now, this week, in addition to the CEO departure from the platform, through TechCrunch, we have been able to know that Elon Musk’s AI is not as guarantor of the truth as its founder says. During the launch of Grok 4, Elon Musk said that Xai’s goal is to create a That seeks the truth to the fullest. However, the first evidence suggests that this search is especially based on the publications and opinions of Musk itself.

An AI that seeks to agree with its head

From what it seems, there are users who have detected that Grok 4 tends to Consult Musk messages in X When you receive questions about delicate issues such as The conflict between Israel and Palestine, abortion or immigration. In addition to various users, TechCrunch also has managed to replicate these results, verifying that the chatbot mentions in its internal reasoning process that seeks Musk’s opinions.

This strategy could be related to Xai’s intention of making Grok less “politically correct”, something that Mussk has criticized publicly. However, it has generated recent controversies, such as anti -Semitic messages sent by Grok’s account in X, which forced the company to rectify and update the chatbot instruction system.

Although Grok 4 tries to offer balanced responses (and probably depending on the Prompt The result is one or the other), it seems that it has been designed to coincide with Musk’s postures. On the other hand, it must be recognized that the model has achieved outstanding technical results in several tests against competition. However, it is useless to have a very powerful AI if you are configured to present a bias so evident in the answers. This endangers a possible massive adoption of Grok 4, either in the business sector through the API or with the subscription of $ 300 per month.