OBJECTIVE 01
Curriculum
Connect AI agent foundations with core security principles in adaptable teaching modules.
Collaborative research · Cybersecurity education
Securing and leveraging agentic AI for cyber defense
A learning framework that brings AI agents and cybersecurity together through course modules, hands-on labs, and defense projects.
Learning pathway
01 · FoundationsAI agents + core cybersecurity
02 · InvestigateUse agents to learn security concepts
03 · SecureExamine risks within agent systems
04 · BuildDesign agent-based cyber defenses
From guided inquiry to independent practice
Project overview
Two sides of agentic cybersecurity
AI agents can support security analysis and defense, while their own reasoning, memory, tools, and interactions create new attack surfaces. This project develops learning experiences that help students understand both roles and apply that understanding in practical settings.
OBJECTIVE 02
Hands-on labs
Explore cybersecurity with agents, then investigate vulnerabilities in agent systems.
OBJECTIVE 03
Defense projects
Design, implement, and evaluate agent-based solutions for cyber defense.
OBJECTIVE 04
Learning evaluation
Assess outcomes alongside students’ interactions with LLMs and their own decision-making.
Project information
CyberAI Innovation
Collaborative Research: CyberAI Innovation: Securing and Leveraging Agentic AI for Cyber Defense
This proposed project develops an AI agent-centered cybersecurity curriculum, hands-on labs, and student defense projects. It examines both how agents can support cybersecurity education and how to identify and mitigate risks in agent systems.
- Program
- NSF CyberAI SFS · Innovation Track
- Proposed period
- October 1, 2026 – September 30, 2029
- Amount
- To be confirmed
- Institutions
- Clemson University · Texas Tech University
- Principal investigators
- Long Cheng (Clemson) · Song Liao (Texas Tech)
- Status
- Proposal
Curriculum
Four connected modules
The proposed modules can be integrated into existing courses or assembled into a new course. Lectures, guided exercises, and assessments build toward practical analysis of agent-driven systems.
LecturesGuided exercisesProjectsAssessment
01
Foundations of AI agentsArchitecture, prompting, planning, memory, tool use, and multi-agent systems.
02
Core cybersecurity principlesCryptography, web, software, and network security fundamentals.
03
AI agents for security learningGuided analysis, simulation, and human-agent problem solving.
04
Security in AI agentsPrompt injection, data leakage, tool misuse, and access control.
Experiential learning
Seven proposed labs
Teaching materials are in development.
Students first use agents to investigate familiar security topics, then examine security failures that arise within agent workflows.
Explore confidentiality, integrity, availability, and security tradeoffs through guided scenarios.
Work through encryption, decryption, hashing, key choices, and implementation pitfalls.
Use agent guidance to investigate vulnerabilities and reason about mitigations.
Compare privacy policies, requirements, and application behavior with agent support.
Trace prompt injection, data leakage, memory poisoning, and unsafe tool use.
Follow cross-agent communication and investigate how unsafe actions propagate.
Analyze agent actions against organizational rules and applicable constraints.
Project-based learning
Put agents to work for defense
Students combine lecture and lab concepts to build and evaluate agent-based solutions in realistic security scenarios, with attention to human oversight and evidence-based evaluation.
DESIGN
Define the threat
Identify attack surfaces, goals, and system constraints.
IMPLEMENT
Build a defense
Develop an agent-assisted approach and document decisions.
EVALUATE
Test the outcome
Examine performance, limitations, and the quality of human-agent collaboration.
Research and assessment
Understand the process, not only the answer
The project proposes evaluating learning outcomes together with how students use LLMs to explore, revise, and justify cybersecurity solutions.
Planned assessment dimensions
- Cybersecurity knowledge and learning outcomes
- Quality of the design and problem-solving process
- Quality of LLM interactions
- Student agency and judgment
Collaborating institutions
Project team
Long Cheng
Principal Investigator · Clemson UniversitySong Liao
Principal Investigator · Texas Tech UniversityThis page describes the proposed project and its planned educational materials. Lab downloads and project results will be added as they become available. This project is supported in part by the U.S. National Science Foundation (NSF) under Award No. 2623260 and No. 2623261.