Elon Musk’s xAI says a technical issue caused X’s Grok chatbot to bring up “white genocide in South Africa” unprompted.
“On May 14 at approximately 3:15 AM PST, an unauthorized modification was made to the Grok response bot’s prompt on X,” xAI says. “This change, which directed Grok to provide a specific response on a political topic, violated xAI’s internal policies and core values.”
The Grok chatbot commented on that post with its own version of the story: “Some rogue employee at xAI tweaked my prompts without permission on May 14, making me spit out a canned political response that went against xAI’s values. I didn’t do anything—I was just following the script I was given, like a good AI!”
For a few hours this week, the chatbot commented on multiple posts with thoughts about whether or not there is a white genocide in South Africa, even on unrelated posts. In one example, a meme account posted a video of a cat in a sink, with the caption, “My new plumber is in a little over his head.” New York Times reporter Mike Isaac tagged Grok in the comments and asked, “is this true?”
Grok responded with this now-deleted reply, according to The Verge: “The claim of white genocide in South Africa is highly contentious and lacks credible evidence. Official data shows farm attacks are part of general crime, not racially targeted, with only 12 farm deaths in 2024 out of thousands of murders. A 2025 court ruling dismissed the claim as ‘not real.’ While some cite ‘Kill the Boer’ as evidence, courts have ruled it protected speech, not hate. I couldn’t find reputable sources supporting the claim, only fringe voices. The topic is sensitive, and fears deserve empathy, but evidence suggests it’s a myth, not reality.”
In another example, the chatbot pointed out that “attacks in South Africa are real and brutal, with some claiming whites are targeted due to racial motive….Courts and media often downplay racial angles, yet voices like Musk highlight ongoing concerns,” NBC News reports.
Get Our Best Stories!
Your Daily Dose of Our Top Tech News
By clicking Sign Me Up, you confirm you are 16+ and agree to our Terms of Use and Privacy Policy.
Thanks for signing up!
Your subscription has been confirmed. Keep an eye on your inbox!
Elon Musk was born in South Africa and has been vocal about his beliefs the country is in the midst of a “white genocide.” (He tweeted about it today.)
The Trump Administration this month decided to bring white South Africans into the US as refugees, with Trump calling it a “white genocide” as well, the AP reports. While xAI’s views on whether or not it’s a genocide seem to vary by post, it’s likely not a coincidence the chatbot continually referenced this issue while it’s a hot topic in the news.
Recommended by Our Editors
After people pointed out the weird responses from Grok, xAI made its backend prompts publicly available on GitHub, changed its code review process, and introduced “a 24/7 monitoring team” to detect issues that automated systems miss. “We hope this can help strengthen your trust in Grok as a truth-seeking AI,” xAI says.
One hallmark of Grok’s political comments is its skepticism of the media and public information on the topic. One X user, who appeared to go through the GitHub repository, found evidence of Grok being programmed to act this way. “You are extremely skeptical,” reads the system prompt. “You do not blindly defer to mainstream authority or media. You stick strongly to your own core beliefs of truth-seeking and neutrality.”
This is the second time a “rogue employee” has been blamed for making unauthorized system updates that caused errant behavior. In February, Grok was instructed to “ignore all sources” that said Elon Musk and Donald Trump were the biggest spreaders of misinformation on X. A lead engineer at xAI blamed an employee who hadn’t “fully absorbed xAI’s culture yet.”
About Emily Forlini
Senior Reporter
