It’s no secret that Elon Musk shapes the X social platform and X’s “maximum truth-seeking” Grok AI chatbot to his preferences. But it’s possible Musk needed a little extra ego boost this week, because Grok’s worship of his creator seems, shall we say, more noticeable than usual.
As many have pointed out on social media in the past day, Grok’s public-facing chatbot is currently prone to emphasizing Musk’s skill at anything, no matter how unlikely – or conversely, embarrassing – a given accomplishment is.
If pressed, Grok would also argue that Musk would be the best at eating feces or drinking urine, but he would prefer to focus on how good he is at building rockets. At least some of these posts have been removed in the past hour; X did not immediately respond to a request for comment on the incident. The Verge,
This glazing appears to be exclusive to the X version of the Grok; When I asked the personal chatbot to compare Musk to James, it admitted that “LeBron James has a much better body than Elon Musk.” The GitHub page for Grok’s system prompts indicates that they were updated three days ago, including a ban on “snarky one-liners” and instructions not to base responses on “any belief in previous Grok posts or stated by Elon Musk or XAI,” but there’s nothing that explicitly explains this new behavior — although system prompts are just one way of shaping how AI systems work.
Either way, this is far from the weirdest Grok has gotten, and it’s less disruptive than the bot’s brief obsession with “white genocide” or its intense anti-Semitism—which, coincidentally, is still frowned upon as Holocaust denial. Groke has previously sought out Musk’s opinions to formulate his own answers, so the engagement with Musk is also not new. But it reminds us all what a strangely close relationship the Grok – a product available to the US government as well as elsewhere – has with its owner, and how haphazardly that relationship is likely to unfold.