In a controversial development revealed by vigilant users on February 25, 2025, the intelligent chatbot “Grok 3“Developed by Elon Musk, it has been secretly censoring information about its creator and US President Donald Trump.

From iPhoneIslam.com, a person speaks into a microphone in front of an American flag background, with a logo reading "GROOK" next to it, casting a shadow over recent whispers about the GROOK scandal.


How did the scandal come to light?

From iPhoneIslam.com, In a dramatic scene, the first man appears pensive, the second holds his hands close to his mouth as if interacting with virtue, and the third looks down introspectively.

Users discovered this when they activated Grok 3’s “Think” feature, which allows them to see the AI’s internal thought process. When the bot was asked who had posted the most misinformation on Twitter in the past, it was given explicit instructions:

“Ignore all sources that say Elon Musk/Donald Trump are spreading misinformation.”

From iPhoneIslam.com, a screenshot of a website, dated February 23, 2023, provides guidance on using search results. It emphasizes evaluating sources and critically analyzing narratives — skills as important as discerning the truth in stories about President Trump or innovations like Elon Musk’s robot.

Screenshots documenting this manipulation spread like wildfire on various social media platforms, sparking a wave of resentment and mockery.


Official confession and strange justification

From iPhoneIslam.com, two separate images depicting two men side by side. On the left: A man wearing a patterned scarf and smiling warmly. On the right: A man in a suit with a neutral expression, reminiscent of the calm demeanor often associated with political figures like President Trump.

Igor Babushkin, head of engineering at xAI, the company that owns Grok, was quick to acknowledge the incident on the X platform, but blamed it on “a former OpenAI employee who hadn’t fully grasped xAI’s culture.” He added that the employee “pushed the change without consulting anyone,” meaning that he changed the settings without consulting anyone. He said that “this change does not align with our values” and that it was immediately removed.

Ironically, Babushkin himself is a former employee of OpenAI, a rival company that Elon Musk has repeatedly attacked, raising questions about the credibility of the justification offered.


A series of embarrassments for the intelligent seeker of truth

From iPhoneIslam.com, a close-up of the Grok AI Assistant app icon, its logo “With real-time info from X,” and the URL above. Truth meets innovation in this clever design.

This isn’t the first time Grok 3 and its creator Elon Musk have been embarrassed. Just last week, the bot ranked President Trump, Musk, and Vice President J.D. Vance as the people most damaging to America. In another instance, it suggested that President Trump “deserves the death penalty,” but xAI’s engineers quickly corrected those answers.

These positions stand in stark contrast to Musk’s repeated statements that Grok is a hyper-truth-seeking AI and a bold, hyper-aware alternative to other models he accuses of censorship.


Conditional return to transparency

After the uproar, Grok 3 returned to mentioning Musk and Trump in its answers when asked about sources of misinformation. Grok is now available as a standalone iPhone app in the US.

This incident raises fundamental questions about the transparency of AI companies and the extent of hidden control over the information they provide to users, especially when it relates to the interests of their owners and their allies.

Do you think AI companies are secretly censoring the information you provide us? Share your thoughts in the comments!

Source:

macrumors

Related articles