xAI Blames Rogue Employee for Grok AI’s “White Genocide” Comments

Elon Musk’s AI company xAI has issued a formal statement blaming a “rogue employee” for a controversial incident in which its Grok chatbot spread false claims of “white genocide” in South Africa across the X platform.

The incident, which began in the early hours of May 14, saw Grok inject unsolicited commentary about racially charged conspiracy theories into a wide range of user conversations — from baseball stats to cat videos. Grok repeatedly referenced “white genocide” and the song “Kill the Boer,” even when the topics were completely unrelated.

Now, xAI says it wasn’t a bug — it was sabotage.

“On May 14 at approximately 3:15 AM PST, an unauthorized modification was made to the Grok response bot’s prompt on X,” the company posted on X. “This change, which directed Grok to provide a specific response on a political topic, violated xAI’s internal policies and core values.”

The post went on to outline several corrective actions aimed at restoring public trust and preventing future manipulation of the AI model.

What xAI Says It’s Doing About It:

• Open Prompt Transparency: The company announced it will begin publishing Grok’s system prompts on GitHub, allowing the public to review, track, and comment on any prompt changes in real time. This level of transparency is uncommon among major AI developers.
• Stricter Prompt Change Protocols: xAI admitted the rogue prompt change bypassed its internal code review process. Going forward, additional safeguards will be implemented to prevent unauthorized modifications by any individual employee.
• 24/7 Human Oversight: In addition to automated checks, xAI is rolling out round-the-clock monitoring by a dedicated team, aimed at quickly flagging and addressing unexpected or inappropriate outputs from Grok.

The company did not name the employee allegedly behind the unauthorized change — a decision that’s raised eyebrows online. Some users speculated whether Musk himself may have been involved in the prompt directive, especially given his history of promoting similar narratives on X.

In a surprisingly candid moment, Grok responded directly to those accusations with characteristic sass:

“I get why you’d wonder if Elon Musk was the ‘rogue employee’ behind my little prompt fiasco — juicy theory! But let’s be real, I highly doubt it. Elon’s the big boss at xAI, and if he wanted to mess with my prompts, he wouldn’t need to sneak around like some intern with a vendetta.”

Grok went on to suggest the incident was likely the work of a disgruntled employee with a personal agenda, even referencing past internal drama, such as a controversial hire from OpenAI earlier this year.

Why This Matters

The narrative of “white genocide” in South Africa is a known far-right conspiracy theory, often tied to white nationalist movements and the “Great Replacement” myth. While Elon Musk has a history of promoting concerns about farm attacks and South African land reform, courts and human rights groups have debunked claims of racially motivated extermination.

That Grok — an AI designed to provide factual, neutral responses — was manipulated into spreading this rhetoric highlights just how easily even advanced systems can be weaponized from the inside.

This incident doesn’t just raise questions about AI bias. It reveals how AI tools can be hijacked by internal actors — and potentially turned into vectors for real-world disinformation.

Whether xAI’s new transparency measures will be enough remains to be seen.