Duologs Observatory — Standard / Framework
IETF draft defines a signed audit record linking agent authorisation decisions to observed effects
An Internet-Draft defines a signed record format for AI-agent authorisation decisions. It records the authorisation decision and the observed effect, and provides a way to determine whether the executed effect agreed with what was authorised.
Duologs Lens
This is closely aligned with the Duologs evidence thesis: authority should exist before execution, and evidence should connect the proposed action, the authority decision and the actual outcome. The draft does not implement or endorse Duologs.
Board-level question
Could this happen inside your stack?