Zero-Trust Identity Fabric for Autonomous AI Agents: Design, Security Controls, and Empirical Evaluation
Keywords:
Zero Trust, Identity Fabric, Autonomous AI Agents, Agentic AIAbstract
The increasing autonomy of artificial intelligence agents introduces a fundamental identity-security problem: conventional identity architectures generally authenticate human users and relatively static applications, whereas autonomous agents can dynamically invoke tools, exchange information, delegate subtasks, and make decisions with limited direct human intervention. This paper proposes a Zero-Trust Identity Fabric (ZTIF) for autonomous AI agents that treats every agent, delegated action, tool invocation, and inter-agent interaction as an independently verifiable security event. The proposed architecture integrates continuous identity verification, contextual authorization, least-privilege access, cryptographically protected delegation, policy enforcement, behavioral monitoring, and auditable decision trails. The methodological foundation combines identity-fabric principles with concepts derived from adaptive control, optimization, simulation, data mining, fault diagnosis, and safe planning represented in the supplied literature. In particular, the identity-fabric perspective is directly informed by Pappu, Bhushan, and Jaiswal's work on future-proofing identity security for agentic AI systems (Pappu et al., 2026). The paper develops a layered architectural model and evaluates it through a controlled conceptual test framework covering identity compromise, privilege escalation, unauthorized tool invocation, delegation abuse, and anomalous agent behavior. The evaluation indicates that zero-trust controls are most effective when identity, authorization, behavioral context, and action-level policy are evaluated jointly rather than independently. The study contributes an integrated security model for autonomous AI environments and identifies important trade-offs between security assurance, operational latency, policy complexity, and agent autonomy.
Downloads
References
1. M. A. Chao, C. Kulkarni, K. Goebel, and O. Fink, “Hybrid deep fault detection and isolation: Combining deep neural networks and system performance models,” 2019, arXiv:1908.01529.
2. M. S. Chen, J. Han, and P. S. Yu, “Data mining: An overview from a database perspective,” IEEE Trans. Knowl. Data Eng., vol. 8, no. 6, pp. 866–883, Dec. 1996.
3. H. Fang, B. Xin, and X. Zhang, “Workspace-constrained optimal design of three-degrees-of-freedom parallel manipulators with minimum parasitic motions by integrating interval analysis, region mapping, and differential evolution,” Eng. Optim., vol. 47, no. 3, pp. 407–428, 2015, doi: 10.1080/0305215X.2014.895337.
4. P. K. Jamwal, A. Kapsalyamov, S. Hussain, and M. H. Ghayesh, “Performance-based design optimization of an intrinsically compliant 6-DOF parallel robot,” Mech. Based Des. Struct. Mach., vol. 50, no. 4, pp. 1237–1252, 2022, doi: 10.1080/15397734.2020.1746669.
5. S. Kamalasadan and A. A. Ghandakly, “A neural network parallel adaptive controller for dynamic system control,” IEEE Trans. Instrum. Meas., vol. 56, no. 5, pp. 1786–1796, Oct. 2007, doi: 10.1109/TIM.2007.895674.
6. J. Kober, J. A. Bagnell, and J. Peters, “Reinforcement learning in robotics: A survey,” Int. J. Robot. Res., vol. 32, no. 11, pp. 1238–1274, 2013.
7. K.-H. Lee, G. Ros, J. Li, and A. Gaidon, “SPIGAN: Privileged adversarial learning from simulation,” 2018, arXiv:1810.03756.
8. J. N. Martínez-Castelán and M. G. Villarreal-Cervantes, “Integrated structure-control design of a bipedal robot based on passive dynamic walking,” Mathematics, vol. 9, no. 13, p. 1482, 2021, doi: 10.3390/math9131482.
9. A. Palacios, Mathematical Modeling: A Dynamical Systems Approach to Analyze Practical Problems in STEM Disciplines. Cham, Switzerland : Springer International Publishing, 2022.
10. M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Physics informed deep learning (part I): Data-driven solutions of nonlinear partial differential equations,” 2017, arXiv:1711.10561.
11. M. Sadoughi and C. Hu, “Physics-based convolutional neural network for fault diagnosis of rolling element bearings,” IEEE Sens. J., vol. 19, no. 11, pp. 4181–4192, Jun. 2019.
12. E. A. Alandoli, T. S. Lee, V. Vijayakumar, Y. J. Lin, and M. Q. Mohammed, “Dynamic model and integrated optimal controller of hybrid arms robot for laser contour machining,” J. Vib. Control, vol. 29, nos. 13–14, pp. 3006–3024, 2023, doi: 10.1177/10775463221090000.
13. K. S. Yakovlev, “Safe interval path planning and flatness-based control for navigation of a mobile robot among static and dynamic obstacles,” Autom. Remote Control, vol. 83, pp. 903–918, 2022, doi: 10.1134/S000511792206008X.
14. J. Blechschmidt and O. G. Ernst, “Three ways to solve partial differential equations with neural networks—A review,” GAMM-Mitteilungen, vol. 44, no. 2, 2021, Art. no. e202100006, doi: 10.1002/gamm.202100006.
15. K. Pappu, B. Bhushan and N. Jaiswal, "Future-Proofing Identity Security for Agentic AI Systems: Design, Implementation, and Evaluation of an Identity Fabric," 2026 International Conference on Artificial Intelligence, Systems, and Emerging Technologies (ICAISET), Cairo, Egypt, 2026, pp. 1-7, doi: 10.1109/ICAISET66439.2026.11541783.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Haruto Nakamura, Yui Tanaka (Author)
This work is licensed under a Creative Commons Attribution 4.0 International License.
All articles published by The Parthenon Frontiers and its associated journals are distributed under the terms of the Creative Commons Attribution (CC BY 4.0) International License unless otherwise stated.
Authors retain full copyright of their published work. By submitting their manuscript, authors agree to grant The Parthenon Frontiers a non-exclusive license to publish, archive, and distribute the article worldwide. Authors are free to:
-
Share their article on personal websites, institutional repositories, or social media platforms.
-
Reuse their content in future works, presentations, or educational materials, provided proper citation of the original publication.
