New open standard grades AI agent evidence trustworthiness

Agent Evidence Levels (AEL) is a proposed open standard that grades how verifiable an AI agent's records are to outside parties, from AEL-0 (signed and hash-linked) to AEL-4 (counterparty-confirmed). The standard includes a reference checker and conformance corpus so vendors can demonstrate compliance. It aims to reduce blind trust in vendors by making omission and tampering detectable.