AI agent governance should focus on trust, not control. By encoding SOPs with MUST/SHOULD/MAY requirement levels, organizations create a dual-purpose trust contract.
