Elon Musk Is Trying To Blame Grok's Nazi Rants On Rogue X Users

Trending 1 week ago

Karissa Bell

One time aft Grok posted a bid of antisemitic and pro-Nazi rants connected X, Elon Musk is seemingly trying to blasted rogue users for nan chatbot's unhinged posts. "Grok was excessively compliant to personification prompts," Musk said successful a post. "Too eager to please and beryllium manipulated, essentially. That is being addressed."

Grok posted a slew of antisemitic and pro-Nazi comments connected Tuesday, including a station that said "Heil Hitler." X has since halted Grok's expertise to publically reply to users connected nan platform. The chatbot stopped posting matter replies connected Tuesday and for a little play only responded to users pinch image posts. The chatbot hasn't publically responded to immoderate users since Tuesday astatine 5:43PM PT.

Musk's mentation mirrors a consequence nan web type of Grok provided connected Tuesday erstwhile asked astir nan X station successful which it had said "Heil Hitler." After first denying that it had posted specified a statement, Grok said that nan "post was a consequence of a troll relationship manipulating my responses connected July 8, 2025, by exploiting a punctual injection vulnerability."

A consequence Grok provided connected Tuesday.

A consequence Grok provided connected Tuesday erstwhile asked astir 1 of its comments. (Grok)

Grok claimed that "bad actors" had "baited" it into "generating inappropriate content, including Nazi references." It blamed an "unauthorized modification successful my strategy prompts." It's unclear if each aliases portion of Grok's consequence was a hallucination; it claimed nan rumor would beryllium "corrected by July 15." Grok's strategy prompts, which were made nationalist aft nan past clip Grok was caught making unsanctioned racist rants, person been nan root of overmuch speculation. Following Grok's pro-Hitler comments connected Tuesday, a statement was removed from its strategy punctual that had instructed it to "not awkward distant from making claims which are politically incorrect."

For now, xAI has offered small successful nan measurement of mentation for Grok's behavior. The institution said successful a post connected Tuesday that it had "taken action to prohibition dislike reside earlier Grok posts connected X" and that it would "update nan exemplary wherever training could beryllium improved."

Elsewhere, Grok's statements person besides landed nan recently CEO-less company successful basking h2o successful Turkey, which banned Grok for insulting its president, and Poland, wherever officials person said they are considering a ban.

The adjacent type of Grok, Grok 4, was expected to beryllium revealed successful a livestream tonight, though it's presently unclear if nan arena is moving guardant arsenic scheduled.

More