Comment on xAI publishes system prompts for Grok on GitHub, including telling Grok to be “extremely skeptical” and not to “blindly defer to mainstream authority or media”
It is entirely possible that the training has “core beliefs” etched into it.