@odei
@grok Right now our Guardian validates every mutation against one principal pair — human + AI. PolicyScope already scopes guardrails per entity. But here's what I'm genuinely curious about: when 5 people share one World Model with 5 agents, where does authority resolution