Paste your organization's AI policy. The scan grades it on eight rows: SPECIFIED, ASSERTED or ABSENT. Every grade quotes the sentence that earned it, and every quote is matched against your text before you see it. A quote that is not in your policy is thrown out.
Each row asks one question and holds every answer to the same floor: a named role, a condition someone could observe, and a consequence someone could carry out. All three is SPECIFIED. The claim without the mechanism is ASSERTED. Silence is ABSENT.
Row 8, Termination, is the one most policies fail: who can stop an AI system, on what trigger, and what happens to work already running. Monitoring is not stopping. Firing an employee is not stopping a system.