Every deployer of high-risk AI is personally carrying unmeasured liability on every inference. The largest AI insurance product in the world — Munich Re's aiSure — writes cover for AI vendors (up to $15M through Mosaic) and for corporations deploying AI at scale, and operates as “soft regulation” through premium signaling. Our inference from the published pages: the affirmative AI products that exist price performance against a metric; none attaches its trigger to a scope the deployer declared before the agent ran.
The actual liability — Article 14 compliance, fleet-level deployer exposure, real-world drift between measurement windows, cascading failures across composed systems — is uninsurable. Not because carriers don't want to write it. Because they cannot measure the thing they need to insure: Verified Role Continuity (VRC) — is the system still doing its authorized role?
Arms-control treaties require independent verification instruments that both parties can read. Without the instrument, the treaty is theater. Without substrate-level VRC, AI insurance is theater. Munich Re built the most sophisticated product possible with existing tools. It can only cover a narrow statistical proxy for the real risk.
The graph: c/t near 1.0, no grounding dimensions. The dot is on the wall. (c/t)^n is the fall rate. n is how many hops separate your model from reality.