A new public domain framework aims to solve the critical problem of AI agent accountability. The Human Root of Trust, created by identity and distributed systems expert Eric Miller, tackles the fundamental assumption that every digital interaction involves a human. This assumption is now broken as AI agents autonomously transact, communicate, and sign contracts, passing human-centric identity checks without any visible human presence.

The framework provides a concrete architecture with three pillars: proof of humanity, hardware-rooted device identity, and action attestation. It outlines a six-step trust chain tracing actions back to a human principal, offering two implementation paths. Miller explicitly dedicates the framework to the public domain, emphasizing no patents, products, or ownership, only the principle of carrying it forward.

The whitepaper is available at humanrootoftrust.org.