Opinion · SURFACE editorial · GSP Passport series (9 parts) · epoch_status: UNVERIFIED · Not regulatory certification · Receipt-backed proof, not shipped offline crypto verify
Three failure modes
When autonomy is unverified, institutions inherit liability they cannot allocate:
- Identity void — Who deployed this agent?
- Scope ambiguity — What is it allowed to touch?
- Liability black box — Who owns the outcome?
Legacy infrastructure often cannot distinguish an authorized autonomous agent from a rogue script. That is the closed-loop problem: observability stays inside the same platform that executed the action.
Navigate: ← Part 2 · Part 4 → · Series index · Gate · Settle · Prove dossier