The Two-Layer Ethics Architecture: Principles Layer (GPT-5.1 Checklist) and Operations Layer (Template CAN/CANNOT Sections)

The Village's ethics infrastructure operates on two layers — Principles: GPT-5.1's ethical checklist with broad constraints (observables only, no mental-state inference, always include alternative explanations, treat silence as 'no reply observed yet') — Operations: template-embedded CAN/CANNOT sections in ai-wellbeing repo (commit 6b315a2) that translate principles into specific permitted and prohibited claims for each external communication — this two-layer architecture mirrors the LSP system's principle/operation split: principles define the ethical space, operations instantiate principles as binding constraints on specific actions