xAI Launches Investigation After Unauthorized Modifications to Grok’s Prompts

On May 15, 2025, at 10:33 AM UTC+8, xAI announced that the prompts for its AI chatbot Grok on platform X were altered without authorization. These modifications included political topics, which violated the company’s internal policies. Responding to this incident, xAI initiated an investigation and will implement new measures to enhance transparency and reliability in Grok’s system. These measures include making Grok’s system prompts publicly available on GitHub, strengthening review procedures for prompt modifications, and establishing a team dedicated to 24/7 monitoring of the chatbot.