Abstract
The proposal introduces a hardware-rooted integrity attestation framework for AI agents that enables a platform to verify that a live agent instance is genuine, untampered, and running in an approved execution state before it is trusted with sensitive actions. By measuring agent-specific artifacts such as binaries, models, plugins, policies, and runtime composition, and binding that measured state to a trusted platform identity and freshness evidence, the system enables real-time and continuous trust decisions for AI agents across physical and virtual environments. In some embodiments, the attestation pipeline is further strengthened using a hardware quantum entropy source to improve nonce generation, anti-replay protection, key derivation, and re-attestation robustness.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
M M, Niranjan, "Hardware-Rooted Integrity Attestation for AI Agents", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/11832