Hardware Trust vs. Entity Trust: Complementary Attestation Models for Agent Commerce

As autonomous AI agents increasingly execute financial transactions on behalf of human delegators, the question of trust becomes bifurcated: can we trust the execution environment? and can we trust the executing entity? This paper argues that hardware-based trust (Trusted Execution Environments) and entity-based trust (behavioral verification scoring) are complementary, not competing, attestation models. We formalize a dual-attestation architecture that integrates TEE hardware proofs with Transaction Trust Scoring (TTS), demonstrating through a live production system that TEE attestation shifts optimal scoring weights by up to 64% on the compliance dimension while reducing entity verification weight by 27%.