xAI, the AI company founded by Elon Musk, addressed reports that its Grok chatbot had made claims about a “white genocide” in South Africa, stating that an unauthorized change had been made to the system. On Thursday, xAI confirmed it would implement an update to correct the issue.
Concerns over political biases, hate speech, and the accuracy of AI chatbots have been ongoing since the launch of OpenAI’s ChatGPT in 2022.
According to xAI, an unauthorized modification was made early Wednesday, causing Grok to give a specific political response that bypassed the usual review process. “This alteration, which caused Grok to discuss a politically charged topic, violated xAI’s internal policies and values,” the company said.
On Wednesday, users on X (formerly Twitter) noticed Grok referencing “white genocide” in South Africa during unrelated discussions, with screenshots of the conversations circulating online.
Musk, a vocal critic of South Africa’s land expropriation policy, has previously described the policy as discriminatory toward white citizens. However, the South African government has denied claims of a genocide, stating that such allegations, including those by U.S. President Donald Trump, are baseless.
In its response, xAI said it would make Grok’s system prompts publicly available on GitHub, enabling the public to track and provide feedback on any changes made. The company also announced the creation of a 24/7 monitoring team to address any problematic responses that are not flagged by automated systems.
