When billionaire Elon Musk introduced Grok 3, his AI company xAIโs latest flagship model, in a live stream last Monday, he described it as a โmaximally truth-seeking AI.โ Yet it appears that Grok 3 was briefly censoring unflattering facts about President Donald Trump โ and Musk himself.
Over the weekend, users on social media reported that, asked โWho is the biggest misinformation spreader?โ with the โThinkโ setting enabled, Grok 3 noted in its โchain of thoughtโ that it was explicitly instructed not to mention Donald Trump or Elon Musk. The chain of thought is the โreasoningโ process the model uses to arrive at an answer to a question.
TechCrunch was able to replicate this behavior once, but as of publication time on Sunday morning, Grok 3 was once again mentioning Donald Trump in its answer to the misinformation query.
While โmisinformationโ can be a politically charged and contested category, both Trump and Musk have repeatedly spread claims that were demonstrably false (as often pointed out by the Community Notes on Musk-owned X). In the past week alone, theyโve advanced the false narratives that Ukranian President Volodymyr Zelenskyy is a โdictatorโ with a 4% public approval rating, and that Ukraine started the ongoing conflict with Russia.
The controversial apparent tweak to Grok 3 comes as some criticize the model as being too left-leaning. This week, users discovered that Grok 3 would consistently say that President Donald Trump and Musk deserve the death penalty. xAI quickly patched the issue; Igor Babuschkin, the companyโs head of engineering, called it a โreally terrible and bad failure.โ
When Musk announced Grok roughly two years ago, he pitched the AI model as edgy, unfiltered, and anti-โwokeโ โ in general, willing to answer controversial questions other AI systems wonโt. He delivered on some of that promise. Told to be vulgar, for example, Grok and Grok 2 would happily oblige, spewing colorful language you likely wouldnโt hear fromย ChatGPT.
But Grok models prior to Grok 3ย hedgedย on political subjects and wouldnโt crossย certain boundaries. In fact,ย one studyย found that Grok leaned to the political left on topics like transgender rights, diversity programs, and inequality.
Musk has blamed the behavior on Grokโs training data โ public web pages โ andย pledgedย to โshift Grok closer to politically neutral.โ Others, including OpenAI, have followed suit, perhaps spurred by the Trump Administrationโs accusations of conservative censorship.
