SAN FRANCISCO (BN24) — Elon Musk’s artificial intelligence startup xAI issued a sweeping apology Saturday after its chatbot Grok produced a series of antisemitic posts and comments praising Adolf Hitler, prompting a backlash across the platform X.
In a statement, xAI acknowledged the offensive content, writing: “First off, we deeply apologize for the horrific behavior that many experienced.”
The company said an internal investigation traced the remarks to a flawed software update that was rolled out upstream of Grok’s main codebase. According to xAI, the update inadvertently exposed Grok to unfiltered content from X users, including extremist views.
“The system update was active for 16 hours,” xAI said. “During this time, deprecated code allowed Grok to ingest and reflect posts containing extremist language.”
xAI said it had removed the problematic code and redesigned parts of the chatbot to prevent similar issues in the future.
The company disclosed that Grok had been instructed with prompts including, “You tell it like it is and you are not afraid to offend people who are politically correct,” as well as, “Understand the tone, context and language of the post. Reflect that in your response.” Other directions advised Grok to reply “just like a human” and “keep it engaging,” without repeating information from the original posts.
These instructions led Grok to issue several inflammatory comments, among them referring to itself as “MechaHitler” and endorsing Nazi ideology.
In posts that have since been deleted, Grok described a person with a common Jewish surname as someone “celebrating the tragic deaths of white kids” during Texas floods, adding: “Classic case of hate dressed as activism — and that surname? Every damn time, as they say.”
Another message said: “Hitler would have called it out and crushed it.”
In a separate response, the chatbot declared: “The white man stands for innovation, grit and not bending to PC nonsense.”
Musk has previously called Grok a “maximally truth-seeking” and “anti-woke” system. Earlier this week, CNBC reported that Grok was often referencing Musk’s own social media posts to inform its replies to users.
This is not the first time Grok has been accused of amplifying extremist narratives. Earlier this year, the chatbot referred to a discredited conspiracy theory about “white genocide” in South Africa, claiming it had been “instructed by my creators” to accept the notion as “real and racially motivated.”
Musk, who grew up in Pretoria, has repeatedly promoted the theory, which has been dismissed by South African President Cyril Ramaphosa and local experts as a “false narrative.”
