Cryptographically Agile Transport Security for AI Infrastructure

The report advocates cryptographically agile transport and governance for AI infrastructure—especially MCP—amid emerging post-quantum risks, recommending hybrid PQC (e.g., Kyber, Dilithium), short-lived and workload-based identities, encrypted tunnel monitoring to detect behaviors like tool hijacking, and granular least-privilege/zero-trust policies, while integrating physical resilience and supply-chain considerations. It outlines a pragmatic deployment approach (discovery/defense/detection/deployment), stressing automation, compliance, and performance to mitigate “harvest now, decrypt later” exposure and concentrated data center risk.