Elon Musk’s xAI faced backlash after users found its AI model, Grok 3, censoring mentions of Donald Trump and Musk in certain responses.

XAI engineering lead Igor Babuschkin confirmed that explicit instructions had caused this issue, which was promptly removed after receiving user feedback.

Media outlets replicated the censorship before xAI fixed it, and by Feb. 23, Grok 3 included Trump in its responses again.

The incident follows earlier concerns that Grok 3 suggested Trump and Musk deserved the death penalty, which xAI called a “terrible failure.”

Musk, who promotes Grok as “anti-woke,” previously acknowledged that earlier versions of the model leaned left due to its training data and pledged to improve its neutrality.

Recent xAI developments