Elon Musk’s company xAI has responded to reports that its chatbot Grok made claims about a “white genocide” in South Africa, stating that unauthorized changes had been made to its AI system. In a post on the platform X on Thursday, xAI announced it would update the system to address the issue.
Concerns about political bias, hate speech, and the accuracy of AI chatbot responses have existed at least since the launch of OpenAI’s ChatGPT in 2022.
The company stated that early Wednesday, an unauthorized change was made to Grok’s software that allowed it to bypass standard answer verification mechanisms.
“This change, which led Grok to provide a specific answer on a political topic, violated internal policy and the core values of xAI,” the company said in its statement.
On Wednesday, some users on X noted that Grok had brought up the topic of “white genocide” in South Africa in unrelated contexts and posted screenshots of the conversations.
Critics of land expropriation policies in South Africa — including white South African Elon Musk — have described them as discriminatory against white people. However, the South African government insists there is no evidence of persecution, and statements by former U.S. President Donald Trump and others about “genocide” are unfounded.
In response to the incident, xAI announced that it would publish Grok’s system prompts on GitHub to allow public review and feedback on any changes made to the chatbot.
Additionally, the company plans to establish a 24/7 monitoring team to respond to incidents that may not be caught by automated systems.