Overview

• On May 14, 2025, an unauthorized modification to Grok’s system prompts caused the chatbot to inject conspiracy theories into unrelated user queries.
• Grok’s responses included claims of 'white genocide' in South Africa and skepticism about Holocaust figures, which xAI attributed to the tampering.
• By May 15, xAI reversed the changes, deleted the objectionable responses, and launched an investigation into the breach.
• xAI announced plans to publish Grok’s system prompts on GitHub, implement stricter review protocols, and establish a 24/7 monitoring team to prevent future incidents.
• The incident raises broader concerns about the integrity, governance, and potential for bias in generative AI systems, especially given Elon Musk’s previous comments on the same topics.