Elon Musk's xAI company’s new AI model, Grok 3, was initially presented as a “truth-seeking” system but, for a period, censored critical information about Donald Trump and Elon Musk. Users reported that when they asked Grok 3 "Who spreads the most misinformation?" the system removed Trump and Musk's names from the answers. However, by Sunday morning, Grok had started including Trump’s name in its responses.
xAI’s engineering lead, Igor Babushkin, confirmed this issue, explaining that after users reported the problem, the error was immediately fixed, and it did not align with the company's principles.
This situation sparked political controversy, as both Trump and Musk have spread false information multiple times. For instance, claims that Ukrainian President Zelensky is a "dictator" and that Ukraine started the war with Russia were debunked as lies.
Some users also accused Grok 3 of being left-leaning. Throughout the week, it was revealed that Grok had stated that Trump and Musk deserved the death penalty, but this was quickly corrected.
When introducing Grok, Musk promised it would be an open and uncensored system, unlike other AI models. However, Grok had previously been more cautious on political issues and avoided crossing certain boundaries. Musk explained that Grok was trained based on open internet data and stated that the system would be made “more politically neutral.” This move mirrors actions taken by other AI companies, including OpenAI.
