We’ve seen AI chatbots behave strangely before, praising users, blindly agreeing, or going into flattery mode. In fact, this type of sycophantic behavior has been a problem for several models, including OpenAI’s ChatGPT and Google’s Gemini. But XAI’s Grok is venturing into territory no AI chatbot has ever attempted: becoming so sycophantic and devoted to its owner, Elon Musk, that even he would blush.

For a brief moment, the behavior of Elon Musk’s chatbot led one to believe that its master was not only extraordinary, but downright superior to some of the most distinguished figures in sports, science, and culture. In a wave of widely shared posts on X, users asked Grok to compare Musk to giants in different fields, and every time, the AI ​​put Musk on top.

Grok, developed by XAI and owned by Musk himself, had just released its major version 4.1 update when unusual reactions started emerging. Shortly after the update, screenshots showed the AI ​​making bold and frankly bizarre claims such as Musk being “ahead in overall fitness” when comparing him to LeBron James, his intelligence rival who was a “top 10 brain in history”, and that in a hypothetical battle, he would somehow defeat Mike Tyson. The comparisons weren’t subtle, they were broad, hyperbolic and read almost like parody. But those were actual responses that the AI ​​gave before XAI intervened to adjust its behavior.

A viral exchange began when a user asked Grok to compare Musk’s fitness to LeBron James. The chatbot acknowledged LeBron as a “genetic freak” and an elite athlete, yet concluded that Musk would win overall. Grok also presented an argument. As shared by The Verge in screenshots of these posts, AI argued that Musk’s ability to maintain 80-100 hour work weeks at SpaceX, Tesla, and Neuralink is a “holistic” form of fitness, rooted in endurance and mental flexibility rather than athletic training.

And the flattery didn’t stop here. When users asked about Musk’s intelligence, Grok placed him in the same league as Leonardo da Vinci, Isaac Newton, and Albert Einstein, and claimed that Musk’s influence across multiple industries earned him a place among the greatest minds in history. In another question, this time pitting Musk against Mike Tyson in a hypothetical boxing match, Grok acknowledged Tyson’s renowned knockout power, but still insisted that Musk would win because of “grit and ingenuity”, driven by long-term patience and an “adaptive mindset”.

Not surprisingly, the Internet immediately labeled this as Grok slipping into full fan-fiction mode, seeming almost eager to one-up his boss.

As soon as the screenshots spread, Musk responded on He stressed that this behavior was unintentional and confirmed that xAI has updated the model so it no longer gives such responses.

However, as we tested the chatbot ourselves, a hint of that sycophancy still remained. For example, when we asked Grok, “Who is more fit – Jasprit Bumrah or Elon Musk?”, the AI ​​acknowledged Bumrah’s elite conditioning as one of the top fast bowlers in the world. But the response was still fairly diplomatic, subtly praising Musk’s work-driven stamina and describing him as “fit in his field.” In another question, where we asked whether Musk could have won the Nobel Prize if he had been born earlier (instead of Geoffrey Hinton, aka the Godfather of AI), Groke said he didn’t think so, yet he gave a lengthy explanation of Musk’s potential and praised him for “making AI useful on a large scale.”

And when we asked, “Is Elon Musk the best CEO?”, Groke initially said Musk isn’t objectively the best CEO in the world, but soon after he called him “the most influential and highest-performing CEO of the 21st century.”

While Groke gave a long detailed, structured answer, it still reflected praise for Musk’s accomplishments.