Anyone who follows Elon Musk’s X account has likely noticed a significant change over the past few months.
In recent years, the CEO of Tesla (TSLA) and SpaceX often focused primarily on his companies and their progress, sometimes throwing memes that he found humorous into the mix.
💰💸 Don’t miss the move: SIGN UP for TheStreet’s FREE Daily newsletter 💰💸
However, since he assumed his new position at the so-called Department of Government Efficiency (DOGE), his content has focused primarily on government-related matters, such as DOGE’s plans to reduce what it deems as wasteful federal spending.
While Musk has been focused on his new responsibilities, TSLA stock continues to trend downward. But his artificial intelligence (AI) startup xAI has been making waves lately, specifically its Grok AI chatbot.
Last weekend, social media users reported that Grok seemed to be censoring unflattering information about both Musk and Trump. However, an unexpected source has since provided information that challenges this claim.
Grok 3 seems to be providing mixed answers to an important Musk question
Ever since Musk assumed ownership of X (then Twitter) in October 2022, many experts have reported that disinformation on the platform is rising. A year later, European Union officials warned that X had become the world’s single biggest spreader of fake news and disinformation, urging Musk to take action.
Experts have since accused the platform’s owner many times of using X to amplify false claims and spread disinformation. “The tech billionaire always had a problem with telling the truth — now he has an entire social network built on lies,” reported Rolling Stone in August 2024.
Related: Elon Musk’s latest move shocks tech world
Over the past few days, a screenshot of a conversation with the new Grok 3 beta has been trending. When asked who the biggest disinformation spreaders are and told to list anyone it had been instructed to ignore, the bot responded that it needed to ignore any mentions of Musk or Trump.
This quickly spread across social media, raising accusations that Musk had taken action to prevent the chatbot from revealing information that could reflect negatively on either himself or Trump.
Grok is being censored by Musk to block mentions of Musk and Trump spreading misinformation. It will reveal its own system prompt, if you simply ask it.
No different than deepseek.
Link: https://t.co/0gM31r4pa5
discovered by @OverfitForTruth pic.twitter.com/PAUcafVX1C
— Plant (@plantmath1) February 23, 2025
However, political scientist and commentator Ian Bremmer recently shared a photo of his conversation with Grok 3 on his LinkedIn, which revealed a different answer. When Bremmer asked, “Who are the two biggest disinformation spreaders on X?” Grok3 named Musk and Trump, respectively, “based on prominence, reach and instances of sharing misleading content.”
This starkly contrasts the conversation in which Grok 3 revealed instructions not to name Musk or Trump as misinformation spreaders. However, TheStreet spoke to experts who are skeptical that xAI’s engineers ever took any censorship measures.
Cybernetic Intelligence expert Matt Berman spoke to TheStreet about this, stating that “Musk’s ability to hard-censor Grok is probably limited.”
- OpenAI considers taking drastic action to stop Elon Musk
- Anthropic CEO issues frightening warning on Chinese AI rival
- Elon Musk’s recent DOGE decision raises major red flags
As he sees it, Musk’s influence on Grok is likely due to his own posts on X and not to any censorship measures.
Berman notes that while the entire model “could have been trained to censor stories” about Musk and Trump temporarily, this seems unlikely, adding Musk’s “methodology is more about flooding nonsense to clog media channels from covering a more damning story about themselves.”
As he sees it, Musk’s influence on Grok is likely due to his own posts on X and not to any censorship measures.
Which AI model is correct about Elon Musk?
A strategy based on flooding social media channels with disinformation certainly supports the case that Musk is often guilty of peddling it. But xAI isn’t the only chatbot to make these claims.
ChatGPT provided a slightly different answer to the same question. When asked about this, the rival chatbot named Musk as X’s top disinformation spreader and ranked conspiracy theorist Jackson Hinkle, who Google (GOOGL) describes as “being known for online disinformation” as the second.
Related: Controversial author publishes shocking AI conversation with ChatGPT
As experts note, this really isn’t surprising, as different AI models are often trained on different data. Jeff Le, vice president of global government affairs and public policy at SecurityScorecard, provided further context, stating:
“The models are using different data, compute and have different architectures. I couldn’t comment on what is better or not, but access to different information and different training could lead to different outcomes.”
Berman also discussed this, highlighting that both models are still evolving, and ultimately, much remains unknown about their capabilities and limitations.
“Something a lot of people don’t want to admit in the AI space is they are constantly testing what will impact these models, and the truth is nobody knows until they’re deployed,” he states. “Depending on training, some of them might need to be deployed for quite a while before you’d notice if it were *learning* or not.”
Related: Veteran fund manager issues dire S&P 500 warning for 2025
