Why it issues: Making a dependable chatbot seems to be the holy grail of AI improvement lately, however as chat interfaces turn into the brand new means folks entry info, the integrity of their responses carries rising societal weight. In response to customers, Elon Musk’s newest AI mannequin referred to as Grok 4 seems to be much less targeted on objectivity and extra for echoing Elon Musk’s erratic on-line persona.
Elon Musk not too long ago launched Grok 4, the most recent iteration of xAI’s basis mannequin, describing it as a “maximally truth-seeking AI.” In response to critics, the mannequin appears to treat Musk himself as the final word supply of reality. His opinions seem to play a central function in Grok’s reasoning course of.
Immediate outcomes shared by customers on X counsel that Grok 4 closely depends on Musk’s posts, public statements, and well-known viewpoints when responding to controversial matters. When requested in regards to the Palestinian genocide by the Israeli military, Grok reportedly supplied the one-word reply “Israel” and could possibly be prompted to justify that reply by referencing Musk’s views. The chatbot was additionally noticed fastidiously analyzing Musk’s statements to raised align with its creator’s considering.
Introducing Grok 4, the world’s strongest AI mannequin. Watch the livestream now: https://t.co/59iDX5s2ck
– xAI (@xai) July 10, 2025
Throughout its launch, Grok 4 was mentioned to point out notable enhancements in a number of complicated exams. Musk claimed it may outperform rival fashions from OpenAI, Google DeepMind, and Anthropic. He additionally emphasised that the chatbot is now much less “woke,” a recurring grievance voiced by the outspoken South African-born billionaire.
Even earlier than the discharge of Grok 4, xAI builders reportedly struggled with the mannequin’s tendency to provide solutions Musk disapproved of. That challenge escalated shortly after July 4, when the chatbot started spouting antisemitic rants in response to person prompts.
I replicated this end result, that Grok focuses practically solely on discovering out what Elon thinks in an effort to align with that, on a recent Grok 4 chat with no customized directions.https://t.co/NgeMpGWBOB https://t.co/MEcrtY3ltR pic.twitter.com/QTWzjtYuxR
– Jeremy Howard (@jeremyphoward) July 10, 2025
At one level, the AI went so far as defining itself the “MechaHitler,” which is actually the least-woke strategy to reply a query. CEO Linda Yaccarino, the general public face of X on the time, stepped down shortly after the incident.
Past antisemitism and the Gaza battle, Grok 4 additionally seems to be consulting Musk’s opinions on different contentious points when wanted, together with immigration in the US. In its common unhinged type, the chatbot makes little effort to hide the sources it depends on. Whereas it often gestures towards a balanced perspective, it virtually at all times finally ends up aligning with Musk’s rhetoric.
When prompted with much less delicate matters, Grok appears to provide extra various responses based mostly by itself reasoning. Nevertheless, the interior workings of the chatbot stay opaque. In contrast to most main AI builders, xAI doesn’t present a system card or technical documentation, leaving customers with few clues about how Grok 4 truly operates.
