Replayable authorization and tamper-evident evidence
Purpose
A normal audit trail can prove that access occurred. Replayable authorization aims to answer whether the access should have been allowed under the rules that applied at the time.
Evidence model
Sensitive retrievals can log a fingerprint of the resolved authorization envelope plus retrieval evidence into tamper-evident history. Permission-bearing state can retain temporal history so membership, roles, locations, and record state can be reconstructed.
Replay modes
- As of then: reconstruct historical rules and compare them with the logged authorization proof.
- As of now: apply current rules to the same historical event to detect permission drift.
AI use
This is especially relevant to questions such as whether an AI system ever received context that was outside the caller’s permitted envelope.