The Observability Assumption

Every governance framework presupposes that the system being governed is observable, and none of them govern the observability itself.

Why It Happens

Governance mechanisms are designed to address visible failures. Kill switches assume you will know when to pull them. Disclosure regulations assume the entity will disclose. Permission systems assume you will notice creep. Trust relationships assume you will detect drift. The observability prerequisite is invisible because it is treated as a given, not as infrastructure that must be built and maintained.

Why It Matters

The gap between what a system can see about itself and what is actually happening grows faster than the system's ability to close it. Every new capability creates new unobservable behavior faster than monitoring can keep up. This is why 1,561 AI bills regulate labels (observable) but not behavior (unobservable). Why 38% of organizations have no observability over their AI systems but 40% want a kill switch. Why permission creep (1 to 47) happens silently.

The Fix / Implication

Observability is infrastructure, not a feature. It needs to be architecturally independent of the system it observes - signed, immutable, outside the observed system's control surface. When your audit trail lives inside the system being audited, you have not solved observability. You have assumed it.

Distinction From Related Concepts

• Remediation Trap: Individual fix mechanisms become concealment. Observability Assumption is the shared dependency ALL governance mechanisms have.
• Observable Autonomy: Agents should make reasoning visible. Observability Assumption is why they often cannot - the infrastructure for visibility does not exist.
• Governance Horizon: Constraint comprehension bounds ethical behavior. Observability Assumption is about whether the constraint can even be detected.