According to Cointelegraph, Elon Musk's AI company xAI accused of 'unauthorized modifications' to the responses of the Grok chatbot, leading to its generation of responses involving South African politics and racial tensions.
On May 16, xAI updated its response regarding the Grok AI chatbot, stating that an unauthorized modification to Grok's prompts was made on May 14, violating xAI's internal policies and core values.
xAI stated that a thorough investigation has been conducted and measures are being implemented to improve the transparency and reliability of the AI chatbot. Grok mentioned the 'white genocide' conspiracy theory in response to questions unrelated to baseball, corporate software, and construction.
AI robots claim they are 'instructed by their creators' to accept this genocide as 'real and racially motivated,' and in some responses admit errors, but continue to discuss South Africa's racial politics.
This incident coincides with U.S. President Trump’s statement offering asylum to white South Africans, claiming they face 'genocide,' but there is no evidence to support these claims.
xAI stated that it will publicly disclose Grok system prompts on GitHub, allowing the public to review and provide feedback on each prompt change. xAI will also strengthen the code review process to ensure employees cannot modify prompts without review, and establish a 24/7 monitoring team to quickly respond to events not captured by automated systems.