This book focuses on agentic AI security, providing a comprehensive guide to the theoretical foundations and practical techniques required to secure the increasingly prevalent AI agent systems. It examines the security challenges posed by multi-agent environments and presents real-world examples of open-source frameworks and commercial solutions to mitigate these risks. It answers key questions, including how to conduct threat modeling for agentic AI systems, how to secure communication and identity within multi-agent environments, and how to leverage open-source frameworks and commercial solutions for effective security.

The book features dedicated chapters on agentic AI threat modeling, identity security, communication security in MAS (Multi-Agent Systems), red teaming, AI agents life cycle security, capability and security benchmarking using GAIA and AIR frameworks, Reinforcement Learning (RL) and security, secure agentic AI deployment strategies, innovative open source security frameworks (Cloud Security Alliance and OWASP examples), and case studies of commercial startups addressing agentic AI security challenges. It also explores the unique threat landscape of agentic AI, the challenges of securing communication and identity within multi-agent systems, and the practical application of security benchmarks and open-source frameworks.

As such, the book equips cybersecurity professionals, AI developers, and researchers with the knowledge and tools to mitigate the unique security risks associated with autonomous agents and multi-agent systems.

Autorentext

Ken Huang is a globally recognized expert in AI and Web3 security, a prolific author, and a leading figure in shaping industry standards. He is the CEO and Chief AI Officer (CAIO) of DistributedApps.ai, specializing in generative AI training and consulting. Ken is deeply involved in driving the development of secure AI systems. He serves as a Research Fellow and Co-Chair of the AI Safety Working Groups at the Cloud Security Alliance (CSA), leading agentic AI initiatives. He is also a Co-Chair of the AI STR Working Group at the World Digital Technology Academy under the UN Framework, and a core contributor to the OWASP GenAI project, focusing on agentic AI security. His expertise extends to his contributions to OWASP's Top 10 Risks report for LLM Applications and his active participation in the NIST Generative AI Public Working Group. He is also a member of the Open AI forum. He has authored and edited numerous influential books in this field. His co-authored book, "Blockchain and Web3: Building the Cryptocurrency, Privacy, and Security Foundations of the Metaverse" (Wiley, 2023), was recognized as a must-read by TechTarget. As a sought-after speaker, Ken has presented at prestigious global forums, including Davos WEF, ACM, IEEE, CSA AI Summit, the Depository Trust & Clearing Corporation (DTCC), and World Bank conferences.

Chris Hughes is the Co-founder and CEO of Aquia, a cybersecurity consulting firm dedicated to securing digital transformation initiatives. With nearly two decades of experience in IT and cybersecurity, Chris leads Aquia with a strong commitment to innovation, security, and impact. He previously served as a Cyber Innovation Fellow (CIF) at the Cybersecurity and Infrastructure Security Agency (CISA), where he focused on advancing software supply chain security. Chris also advises several emerging technology startups in areas including Software Composition Analysis (SCA), Kubernetes Security, Non-Human Identities (NHI), and AI Security. In the private sector, Chris has worked as a consultant and currently serves as an adjunct professor for cybersecurity master s programs at the University of Maryland Global Campus. He actively contributes to the cybersecurity community through his involvement in industry groups such as the Cloud Security Alliance s Incident Response and SaaS Security Working Groups, and serves as the Membership Chair for Cloud Security Alliance D.C..

Inhalt

Part 1: Foundations of Agentic AI Security.- Introduction to Agentic AI: Foundations, Drivers, and Risks.- Agentic AI Threat Modeling.- Agentic AI Identity Security.- Agentic AI Communication Protocols and Security.- Part 2: Advanced Security Techniques and Frameworks.- AI Agents Life Cycle and Security Considerations.- Agentic AI Capability and Security Benchmark.- Agentic AI Reinforcement Learning and Security.- Agentic AI Red Teaming.- Securing Multi-Modal Agentic AI Systems.- Part 3: Guidance on Real-World Deployment and Commercial Landscape of Agentic AI Security.- Deploying Agentic AI in Enterprise Environments.- Agentic AI Security in Critical Sectors Finance, Healthcare, and Autonomous Driving.- The Commercial Landscape of Agentic AI Security.