A model's weights can be altered after deployment. Most security infrastructure around AI does not detect this. CBIA is a monitoring system that watches a deployed model from the outside, with no access to its internals, continuously probing its behavior and flagging statistically significant deviations. Every verdict is cryptographically signed and chained to the one before it. As AI systems take on regulated, high-stakes roles in medicine, finance, and infrastructure, the absence of this kind of continuous behavioral attestation is a liability waiting to be named.