Cici: ucss: securing research cyberinfrastructures with zero trust and agentic AI

The National Science Foundation Office of Advanced Cyberinfrastructure awarded Texas A&M University-Corpus Christi $599,958 on August 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop ZT-Agents, artificial intelligence-powered cybersecurity solutions for research cyberinfrastructures. The project runs through July 31, 2029, with performance in Corpus Christi, Texas.

ZT-Agents leverage agentic AI techniques to provide next-generation cyber-protections by enabling continuous evaluation and enforcement of authorization policies for research computing infrastructure. The work addresses the challenge that research infrastructure administrators, researchers, and students struggle to correctly write, evaluate, and enforce the varied authorization policies needed for effective cyber-protections under a zero-trust paradigm. The project advances state-of-the-art defenses through AI agents for continuous authorization, supports practitioners in learning and retrofitting agentic AI-augmented authorization technologies that balance security and usability, develops evidence-backed technical recommendations and best practices for zero-trust continuous authorization, and disseminates cybersecurity and agentic AI solutions through hands-on challenges to K-12 and college students and research infrastructure practitioners via publications and other outlets.

Abstract

Research cyberinfrastructure (RCI) supports cutting-edge scientific projects benefitting individuals, the environment, national security, and economic competitiveness. Despite the potential benefits, recent attacks have compromised the integrity of scientific data and resources such as computing time and network bandwidth, ultimately affecting the public?s confidence in scientific results. As a response, countermeasures have been proposed based on zero trust (ZT), an emerging paradigm calling for the continuous evaluation and enforcement of authorization policies, which protect rcis by restricting access to sensitive data and resources. However, IT is difficult for practitioners of rcis, namely, administrators, researchers, and students, to correctly write, evaluate, and enforce the many different authorization policies needed to provide effective cyber-protections. To address these challenges, this project develops ZT-agents, which leverage agentic artificial intelligence (AI) techniques to provide next-generation cyber-protections for rcis, ultimately pursuing the following goals: (i) advancing the state-of-the-art in cyber-protections by developing AI agents for continuous evaluation and enforcement of authorization policies for rcis; (II) enabling practitioners to conveniently learn and retrofit agentic AI-augmented authorization technologies for rcis, balancing security, usability, and the needs of research-driven enterprises; (III) providing evidence-backed technical recommendations and best practices on ZT-inspired continuous authorization technologies for rcis, thus favoring efficient implementations in practice; (IV) disseminating next-generation cybersecurity and agentic AI solutions via hands-on challenges to students in college and k-12 levels, as well as to practitioners of rcis; and, (v) disseminating the results of this project via publications in well-reputed scientific venues as well as articles and presentations for public audiences. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are planned for this award.