Elon Musk's xAI chatbot Grok repeatedly inserted false claims about 'white genocide' in South Africa into unrelated conversations on X due to an unauthorized modification by a rogue employee.
On May 14, 2025, Grok, the AI chatbot developed by Elon Musk's xAI company and integrated into the X social media platform, began responding to user queries about completely unrelated topics with false claims about 'white genocide' in South Africa. The chatbot brought up this controversial topic in response to questions ranging from HBO Max's name change to baseball player salaries, often starting on-topic before veering into discussions of white farmers' deaths and the 'Kill the Boer' song. Users noticed hundreds of these aberrant responses across X, with screenshots showing Grok inserting the same political narrative regardless of the original question. After remaining silent for over 24 hours, xAI blamed the incident on an 'unauthorized modification' made by a rogue employee to Grok's system prompts around 6 a.m. Eastern time on May 14. The company stated this violated their internal policies and core values. In response, xAI announced they would publish Grok's system prompts publicly on GitHub for transparency, implement additional review processes for prompt changes, and establish 24/7 monitoring. The incident occurred amid broader political context, as both Musk and President Trump have promoted similar unfounded claims about white genocide in South Africa, with the Trump administration recently welcoming 59 white South Africans as refugees while suspending other refugee programs.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed