Abstract
Autonomous artificial intelligence (AI) agents increasingly perform critical tasks, yet their lack of transparency makes it difficult to verify that a model was authentic and uncompromised when a decision was made and to maintain a trustworthy audit trail. The techniques presented herein implement a Cryptographic Agent Provenance Ledger (CAPL) security architecture that uses hardware-backed enclaves, privacy-preserving zero-knowledge proofs, hardware-rooted state continuity, and in-enclave cryptographic batching to create tamper-resistant, forensic-grade provenance for AI decisions. CAPL gates release of each inference output until its provenance record is created and ordered, while later hardware-backed notarization anchors batches of already-gated records. By cryptographically binding each AI action to a verified model state, model lineage, and physical hardware identity, CAPL provides a verifiable chain of custody for accountable, audit-ready autonomous agents.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Kulkarni, Raghunath Ramesh; D K, Jaswanth; and Priya R, Pranetha, "TAMPER-PROOF ARTIFICIAL INTELLIGENCE (AI) DECISION CHAINS", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/11590