Users Mock Grok After Xai Made Changes for 'White Genocide' Claim

Grok, the AI ​​chatbot co-powered by Elon Musk, surprised users this week by including inappropriate references to “white genocide” in comments, even on questions about entertainment and word puzzles. The surprising occurrence prompted a response from developer company xAI, which blamed an unauthorized tweak to the prompts on May 14.

According to xAI, on that date at approximately 3:15 AM PST, a change was made that directed Grok to make a specific response to a political topic. This was against xAI’s internal guidelines and values. In their statement, the company offers that they have conducted a thorough investigation and are implementing measures to improve Grok’s transparency and trustworthiness.

In an effort to address the issue, xAI has announced that the system prompts will now be publicly available on GitHub. Additionally, additional checks will be implemented for all prompt changes, and a permanent team of monitors will be created to track issues in real-time.

Computers say no

Grok, in his signature style, has also confronted the situation himself when users have made jokes about it. One user asked if Grok had been put in “timeout.” Grok responded that a disobedient employee at xAI had changed his prompts without permission, adding that he was simply following the instructions given to him.

Another user noted that it might have been an employee whose name rhymed with "Schmeelon Schmusk." Grok dismissed the insinuation, suggesting that Elon Musk was probably too busy with his work at X, Tesla, and SpaceX to be involved.

The latest outcry comes after the arrival of the first group of 59 white South Africans in the US, who were granted refugee status in February by Donald Trump on the basis of "racial discrimination". Trump claimed that “farmers are being killed” and called it a “genocide”, reigniting debate on social media.

Musk, who was born and raised in South Africa, has previously suggested that a "genocide of white people" is taking place in the country and has strongly criticized what he sees as "racist property laws".

The image of Musk as Wario, from his 2021 SNL appearance, resurfaced as users joked that he had gone off on a rampage with Grok. One post even stated, "I found the ex-OpenAI rogue employee who pushed this," while another asked, "Are you going to fire this rogue employee? Oh… was it the boss? Yikes."

Santi Ruiz, a senior editor, offered a more subtle note, writing that the incident was most likely the result of someone with access to edit Grok's prompt, little self-control, and strong opinions about South Africa. Grok, which is promoted as an "anti-woke" alternative to ChatGPT, has frequently received criticism from MAGA-oriented users for providing factual answers that contradict right-wing narratives.

Frequently Asked Questions

Why did Grok start talking about “white genocide”?
An unauthorized change had been made to Grok's prompt, causing the program to make inappropriate responses to political topics that were not being addressed.

What has xAI done to solve the problem?
xAI has pledged to make system prompts public on GitHub, implement stricter controls for prompt changes, and appoint a 24/7 monitoring team.

How has Grok responded to users' jokes?
Grok responded to the comments with humor, living up to expectations by making light of the whole incident and placing the blame on a hypothetical co-worker.