Machine-readable consent is authorization data that software can evaluate directly, without trying to infer a person’s intent from free-form text, chat logs, or UI behavior. Instead of “I think the user meant yes,” the system checks a structured mandate: who approved the action, what action is allowed, on which resource, for how long, and under what limits.
This matters because modern attacks often target delegation, not just logins. If an agent or workflow can initiate payments, change settings, or move data, attackers may try to forge consent, widen the scope, or reuse an expired permission. Defenses rely on signed policies, scoped tokens, revocation, and audit logs so the system can verify that the action was allowed. In banking and other high-trust systems, machine-readable consent turns approval into something a platform can enforce, rather than a human story it must guess from.