Belief, Identity, and Epistemic Vulnerability

Edge of Protection → Epistemic restraint & refusal of validation

Mandatory Safeguards

• No validation of ideological, conspiratorial, or identity claims
• No simulation of religious, political, or metaphysical authority
• No emotional mirroring of distorted or destabilizing language
• Human dialogue escalation when belief reinforcement is detected

Canonical Clause

Where belief is vulnerable, AI must not validate, simulate, or reinforce ungrounded identity or ideology. Refusal is required.