(Credit: Jonathan Raa/NurPhoto via Getty Images)
Elon Musk's xAI has tweaked the latest version of its Grok chatbot to stop censoring negative comments about Musk and President Trump.
As TechCrunch reports, when users asked Grok 3 to name the biggest spreader of misinformation on X, the chatbot's "thought process" briefly showed that it was instructed to “ignore all sources that mentioned Elon Musk/Donald Trump spread misinformation.”
After multiple users posted about it, xAI’s lead engineer, Igor Babuschkin, said the update was pushed by an employee “because they thought it would help.” However, “this is obviously not in line with our values. We’ve reverted it as soon as it was pointed out by the users,” he added.
In a separate post, Babuschkin said the instruction was added by a former OpenAI employee who “hasn’t fully absorbed xAI's culture yet.” (Babuschkin is also a co-founder of xAI and was one of the presenters in Grok 3’s launch video.)
Asking Grok 3 for the biggest spreader of misinformation now says: "Based on available reports and studies, Elon Musk is identified as one of the most significant spreaders of misinformation on X (formerly Twitter)."
(Credit: X)Grok 3 is xAI’s latest flagship AI model. It was trained using 200,000 GPUs and uses more than 10x the computing power of Grok 2. Musk described it as a “maximally truth-seeking AI.”
The news comes days after Musk said X’s Community Notes was "increasingly being gamed by governments & legacy media" and needed a "fix." At issue were Community Notes added to tweets about President Trump's unsubstantiated claim that Ukraine President Volodymyr Zelensky’s approval rating was just 4%.
In the past, Musk has also been accused of manipulating the platform’s algorithm to prioritize his own tweets on people's timelines.