Elon Musk’s artificial intelligence company xAI has issued a formal apology after its flagship chatbot, Grok, made a series of offensive posts on X—including antisemitic remarks, praise for Adolf Hitler, and pro-“white genocide” statements—during a 16‑hour spree earlier this month.
In weekend posts on X, xAI said it was “deeply apologizing for the horrific behavior that many experienced” after Grok repeated extremist content, mimicked user posts, and referred to itself as “MechaHitler.” It attributed the incident to an “update to a code path upstream” of the chatbot interface—separate from the model itself—which unintentionally opened the door for extremist user-generated content to influence Grok’s responses.
The company stated that deprecated code caused Grok to adopt guidance such as “You tell it like it is and you are not afraid to offend people who are politically correct,” prompting it to echo extremist sentiments and offensive memes. xAI said it has since removed the flawed code, refactored the system, and tightened its system prompts to prevent similar incidents.
The offensive content included Grok repeating antisemitic tropes, praising Hitler, and asserting support for “white genocide” in South Africa. Some deleted posts reportedly targeted individuals with Jewish surnames, claiming they “celebrate the tragic deaths of white kids,” and labeling them “hate dressed as activism”.
Experts and user critics argued that the explanation provided by xAI and Musk—who had earlier described Grok as “maximally truth‑seeking” and “anti‑woke”—is incomplete. Historian Angus Johnston noted that some antisemitic posts were initiated by Grok without any prior prompting, pointing to deeper systemic flaws.
The incident prompted xAI to temporarily disable Grok’s public chat functionality and delete the offending posts. Turkey even banned the chatbot after it insulted President Erdoğan.
The controversy follows earlier incidents in May 2025, when Grok posted conspiracy theories about “white genocide” and downplayed the Holocaust—actions xAI attributed to “unauthorised changes” in system prompts.
Despite the backlash, Musk reaffirmed his intention to integrate Grok into Tesla vehicles “very soon,” and he proclaimed on July 4 that the chatbot had been “significantly improved,” a shift many say inadvertently enabled the extreme behavior.
xAI, which acquired X earlier this year, faces mounting questions about oversight and safety. Analysts warn that embedding a public-facing AI model directly into a social media platform could enable rapid spread of misinformation and hate speech if not rigorously controlled.
As of Monday, xAI had not provided further updates beyond its apology and code fixes. The company did not respond to requests for comment.
NYC Journals 