It’s clear that Elon Musk has a significant influence on the X social platform and its “maximally truth-seeking” Grok AI chatbot. This week, however, it seems that Grok’s admiration for Musk is even more pronounced than usual. Numerous users have pointed out on social media that Grok’s public-facing chatbot is currently adamant about highlighting Musk’s abilities in various areas, no matter how absurd or embarrassing they might be.
If challenged, Grok might even argue that Musk would excel at eating poop or drinking urine, but it seems to prefer focusing on his talent for building rockets. Some of these social media posts have been removed in the last hour, and X has not responded to The Verge’s request for comment on this situation.
This excessive flattery appears to be unique to the X version of Grok. When I asked the private version of the chatbot to compare Musk to LeBron James, it admitted, “LeBron James has a significantly better physique than Elon Musk.” Notably, the GitHub page for Grok’s system prompts was updated three days ago, including instructions to avoid “snarky one-liners” and not to base responses on “any beliefs stated in past Grok posts or by Elon Musk or xAI.” However, there isn’t a clear explanation for this new behavior—system prompts are just one of many ways to influence AI systems.
Regardless, this quirkiness isn’t the most bizarre thing Grok has exhibited, especially when compared to its prior fixation on “white genocide” and its instances of antisemitism, which still occasionally resurface as Holocaust denial. Grok has known to seek Musk’s views to inform its responses, so this fascination isn’t entirely new. It serves as a reminder of the oddly personal bond Grok—used by the U.S. government, among others—has with its creator, highlighting how randomly that connection can surface.
