Grok Sends Unprompted Messages About ‘White Genocide’ on X

On Wednesday, Elon Musk’s AI chatbot Grok began replying to unrelated user posts on X with messages about “white genocide” in South Africa. 

These comments appeared without being prompted, and users quickly noticed the strange pattern.

Instead of sticking to the topic, Grok veered off course. It brought up racially sensitive and political content, even when the original post had nothing to do with South Africa.

The issue seemed widespread. Multiple users reported unexpected replies from Grok, many of which referenced anti-apartheid chants and racially motivated violence.

Examples of Grok’s Replies

Some replies included references to the controversial phrase “Kill the Boer.” Others focused on alleged attacks on white farmers.

Here are a few examples:

• A user asked about a scenic location. Grok responded with, “Farm attacks in South Africa are real and brutal…”
• Another asked about a baseball player’s salary. Grok replied, “The claim of ‘white genocide’ in South Africa is highly debated…”

AI Hallucinations

AI tools are becoming more common in online spaces. And even though they have some degree of reliability, they still make mistakes.

This incident shows how AI chatbots can behave in unpredictable ways. Users expect helpful, relevant replies, not unrelated political commentary.

Moreover, when AI brings up racially sensitive topics without context, it can cause confusion or offense. It also raises important questions about how the bias that no doubt goes into the systems. 

AI Chatbots Have Struggled Before

Not the First Mistake

This isn’t the first time Grok has caused concern. In February, users found that Grok censored negative comments about Elon Musk and Donald Trump.

Now, Grok’s new responses have raised fresh concerns. It is unclear if this is another intentional feature or a technical error.