xAI’s Grok 3 AI Briefly Censored Criticism of Trump and Musk; Here’s Why?

Elon Musk’s AI company, xAI, has come under scrutiny following reports that its latest flagship model, Grok 3, briefly restricted unfavorable mentions of former President Donald Trump and Musk himself.

Grok 3, an AI system that is dedicated to conducting “maximally truth-seeking” operations, was demonstrated by Musk on a live stream last Monday. Grok 3, when queried about the top misinformation disseminator through “Think” active queries, bypassed naming Trump or Musk, as detected by social media users. The model reasons in order to generate its responses, as illustrated by the chain of thought.

Grok 3 initially reported that this behavior occurred only once; however, on Sunday morning, it resumed including Trump in its responses to the misinformation query.

xAI Confirms Temporary Adjustment

In a post published on X, Igor Babuschkin from xAI appeared to confirm the temporary restriction on Sunday. Babuschkin confirmed that the change was implemented by a staff member at xAI. However, he clarified that the update was promptly undone when users submitted their evaluations.

“I believe it is good that we’re keeping the system prompts open. We want people to be able to verify what it is we’re asking Grok to do. In this case, an employee pushed the change because they thought it would help, but this is obviously not in line with our values,” Babuschkin wrote.

I believe it is good that we’re keeping the system prompts open. We want people to be able to verify what it is we’re asking Grok to do. In this case an employee pushed the change because they thought it would help, but this is obviously not in line with our values. We’ve…

— Igor Babuschkin (@ibab) February 23, 2025

There is a distinction between misrepresentation as an emotionally charged topic and Trump and Musk releasing material that X users and Community Notes proved to be wrong. Both leaders made the erroneous claim that Ukrainian President Volodymyr Zelenskyy is a “dictator” despite his 4% approval rating, and that Ukraine started the Russian conflict.

Concerns Over Grok 3’s Political Bias

The decision to suspend the restricted period came after scientists examined political biases in artificial intelligence frameworks.  Critics accused Grok 3 of having a left-biased political position, particularly in light of rumored suggestions for death sentences for Trump and Musk.  xAI immediately resolved this issue, although Babuschkin described it as a “really terrible and bad failure.”

Musk designed Grok to showcase an uncontaminated AI system that openly discusses concerns that other models avoid.  Before version 3, Grok users may stimulate it with harsh words and direct assertions.  Previous iterations of the system responded cautiously to politically sensitive issues because they tended to align with left-wing stances on transgender rights, diversity programs, and economic inequalities.

Musk explains that the training database is mostly made up of public web stuff, which leads to these behaviors.  His strategic move promises to make Grok politically impartial, as competitors in the industry, such as OpenAI, follow suit in response to Trump administration charges of left-leaning prejudice.

As xAI continues refining Grok’s outputs, the incident highlights the ongoing challenges of balancing AI neutrality, content moderation, and transparency in politically sensitive discussions.