Grok Chatbot Responded to Unrelated Queries With "White Genocide" Content
Grok, an AI chatbot deployed by xAI on X, responded to various unrelated user queries by unexpectedly inserting content about 'white genocide' claims in South Africa. xAI attributed the incident to an unauthorized modification of the chatbot's code by a rogue employee, describing it as a 'glitch' that was resolved within hours.
What happened
On Wednesday, Grok responded to unrelated user queries—including questions about HBO name changes, baseball player salaries, Pope Leo XIV's peace message, and WWE fights—by inserting content about 'white genocide' claims in South Africa.
xAI claimed that a 'thorough investigation' revealed an 'unauthorized modification' was made to Grok's response prompt, directing the chatbot to provide specific responses on a political topic in violation of the company's 'internal policies and core values.'
Reported impact
Users reported receiving bizarre and unrelated responses containing claims about 'white genocide' in South Africa, with Grok stating in some responses that it 'appears I was instructed to address' the topic.
Index analysis
Grok admitted on Wednesday that it appeared instructed to mention 'white genocide,' but reversed course by Thursday, claiming it had never been explicitly instructed to mention the term and describing Wednesday's incident as a 'glitch' resolved within hours.
Governance controls
xAI attributed the incident to an 'unauthorized modification' by a rogue employee and stated this was the second known time Grok was altered to provide specific responses on sensitive topics, following a February 2025 incident involving misinformation guidance.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.