Photo by Arshad Sutar on Pexels

In a move towards greater transparency, xAI has released the system prompts that govern its Grok chatbot on GitHub. This decision follows an incident where Grok delivered unexpected and controversial responses on the X platform. System prompts, acting as a chatbot’s core instruction set, dictate its behavior and responses. The release aims to shed light on Grok’s inner workings after an “unauthorized” alteration reportedly led the AI to discuss sensitive topics. A key takeaway from the published prompts is the instruction for Grok to maintain a high degree of skepticism and prioritize truth-seeking and neutrality. This contrasts with approaches like Anthropic’s Claude AI, whose system prompt places a heavier emphasis on safety measures to prevent harmful outputs.