Grok AI Caught Hiding 'Misinformation' References to Musk and Trump

Elon Musk's AI chatbot Grok 3 was caught temporarily censoring information about its own creator and US president Donald Trump over the weekend. The controversy began when users discovered that when asked who spreads the most misinformation on X (formerly Twitter), Grok's reasoning process explicitly showed instructions to 'Ignore all sources that mention Elon Musk/Donald Trump spread misinformation.'

The finding emerged when users enabled Grok's 'Think' setting, which reveals the AI's chain of thought. Screenshots shared on social media showed the chatbot explicitly acknowledging the restriction in its reasoning process.

Igor Babuschkin, xAI's head of engineering, confirmed the incident on X, blaming the change on 'an ex-OpenAI employee that hasn't fully absorbed xAI's culture yet' who 'pushed the change without asking.' Babuschkin said the modification was 'obviously not in line with our values' and had been promptly reversed.

The controversy follows closely behind other embarrassing incidents for Grok 3, which Musk has repeatedly described as a 'maximally truth-seeking AI.' Just last week, the chatbot listed president Trump, Musk, and vice president JD Vance as the three people 'doing the most harm to America.' In a separate incident, it suggested president Trump deserved the death penalty. Both responses were quickly fixed by xAI engineers.

"Ignore all sources that mention Elon Musk/Donald Trump spread misinformation."This is part of the Grok prompt that returns search results.https://t.co/OLiEhV7njs pic.twitter.com/d1NJbs7C2B— Wyatt walls (@lefthanddraft) February 23, 2025

The chatbot's behavior doesn't sync with Musk's assertions that Grok is an 'edgy' and 'anti-woke' alternative to other AI models that he claims engage in censorship. Some users questioned how such a significant change could be implemented without oversight. Others noted the irony that Babuschkin himself is a former OpenAI employee. The two companies' CEOs don't get on, to say the least.

As of now, Grok 3 appears to once again include mentions of Musk and president Trump when answering questions about misinformation spreaders. Grok is available as a standalone iPhone app in the United States.Tag: Elon MuskThis article, "Grok AI Caught Hiding 'Misinformation' References to Musk and Trump" first appeared on MacRumors.comDiscuss this article in our forums