Securing Generative AI, LLMs, and ML using Zero Trust Architecture

Securing AI Using Zero Trust Principles

Strategic Guidance for Defending AI Systems in a Rapidly Evolving Threat Landscape

Artificial intelligence is reshaping industries, driving innovation in critical sectors such as healthcare, finance, energy, and government. Yet, as organizations integrate AI into business operations, they inherit new risks, many of which conventional security models fail to address. Adversaries are weaponizing AI to automate reconnaissance, bypass defenses, and exploit vulnerable systems. The solution is not more trust, but less.

Zero Trust offers a foundational paradigm shift: no identity, device, system, or interaction is inherently trusted. Security must be continuously enforced, context-aware, and resilient by design. This book demonstrates how Zero Trust, when strategically applied to AI environments, enables organizations to secure data pipelines, mitigate emergent threats, and maintain control over evolving digital ecosystems.

Key insights include

• AI Through a Security Lens: Demystifies machine learning, generative AI, and large language models with a focus on operational and business impact.
• Zero Trust Foundations: Provides a historical and architectural overview of Zero Trust, including Cisco's Five Zero Trust Categories.
• Security by Design for AI: Offers guidance on protecting AI development workflows, from data ingestion and model training to inference and deployment.
• Threat Mitigation Strategies: Addresses adversarial AI, data poisoning, shadow AI, and insider misuse through identity enforcement, segmentation, and telemetry.
• Strategic Execution: Maps Zero Trust principles to regulatory frameworks including NIST AI RMF, EU AI Act, DORA, and ISO 27001, and provides actionable templates for running successful Zero Trust Segmentation Workshops.

Who Should Read This Book:

• CISOs and security architects building AI-resilient architectures
• AI and data leaders embedding AI into enterprise infrastructure
• Risk, compliance, and governance professionals navigating regulatory change
• Technical teams seeking secure-by-design methodologies for AI initiatives

Why This Matters Now:

AI systems are expanding faster than most organizations can govern them. The risks, ranging from operational disruption to model corruption, require proactive, architectural defenses. This book bridges the gap between AI innovation and trusted enterprise security.

Securing AI Using Zero Trust Principles delivers the strategic playbook for building resilient, trustworthy, and standards-aligned AI systems that can withstand the threats of today and tomorrow.

Cindy Green-Ortiz is a globally recognized cybersecurity strategist, principal architect, and trusted advisor to Fortune 100 enterprises and public sector leaders. With 40 years of experience in security and technology leadership, Cindy has guided complex organizations across industries--including financial services, healthcare, hospitality, military, energy, and manufacturing--in transforming their security posture and achieving sustainable business outcomes.

At Cisco, Cindy is a principal security architect, leading global Zero Trust initiatives. She co-led Cisco's AI program, Wintermute, and co-led the Post Quantum Resistant Cryptography Working Group. As a Cisco Press author, Cindy translates technical depth into practical frameworks that inspire engineers, architects, and executive leaders alike. Her ability to bridge cutting-edge technology with security management has made her a sought-after public speaker and educator.

Cindy has delivered impactful presentations and workshops at Cisco Live, Offensive Summit, Cisco Secure, ISC2, ISACA, and WiCyS, engaging audiences from the boardroom to the classroom. She is committed to mentoring the next generation of cybersecurity professionals and frequently collaborates with academic institutions to advance cybersecurity education, research, and policy.

Publications:

--Cisco Live Speaker: BRKXAR-2008: Navigating the Future of Cybersecurity: AI, Quantum-Resistant Cryptography, and Zero Trust (2025); BRKXAR-2008: Exploring the Paradigm Shift in Security: AI and Quantum Cryptography's Influence on Zero Trust (2025)

--Co-Author: Zero Trust Architecture (2023)

--Technical Editor: In Zero Trust We Trust (2024)

--Cisco Insider Advocate: "Ask Me Anything" (2023)

--Co-Author of a Cisco Whitepaper: "IPv6 Addressing Analysis for the US Army" (Clearance required, 2023)

--Cisco Live Speaker: "BRKXAR-2008 Zero Trust Segmentation" (2022-2024)

--DHS Whitepapers: "Going Dark: Impacts of Encryption" (2017); "Digital Blackmail (Ransomware) as an Emerging Tactic" (2016); Wang Laboratories, Chairman's Golden Circle

Zig Zsiga, CCDE 2016::32, CCIE #44883, CISSP, has been in the networking industry for 20 years. He is currently a principal architect supporting the Cisco CX U.S. public sector business and customers. Zig holds an active CCDE and two CCIE certifications, one in Routing and Switching and the second in Service Provider. He also holds a BS in computer science from Park University. He is a father, a husband, a United States Marine, a gamer, a nerd, a geek, and a big soccer fan. Zig loves all technology and can usually be found in the lab learning and teaching others. This is his second published book, and he is also the host of the Zigbits Network Design Podcast (ZNDP), where he interviews leading industry experts about network design. All of Zig's content is located at https: //zigbits.tech.

Publications:

--Author: Cisco Certified Design Expert (CCDE 400-007) Official Cert Guide (2023)

--Cisco Insider Advocate: "Ask Me Anything" (2023)

--Cisco Live Speaker:

--BRKRST-2044: Enterprise Multi-Homed Internet Edge Architectures (2017-Present)

--LTRENT-2016: Learning IPv6 in the Enterprise for Fun and (Fake) Profit: A Hands-On Lab (2018-Present)

--TECCRT-3005: CCDE Techtorial (2019-Present)

--LTRCRT-3000: CCDE Practical Exam Practice Lab (2020-Present)

--LTRENT-2016: Learning VxLAN in the Enterprise for Fun and (Fake) Profit: A Hands-On Lab (2024-Present)

Saskia Laura Schröer holds a PhD with a specialization in Information Systems and an MSc focusing on data science. Saskia is certified in CCNP Enterprise and is a DevNet Associate, a WiCyS speaker, and a Cisco Live speaker. Saskia is a security consulting engineer in Cisco's EMEA Cybersecurity Centre of Excellence with almost 10 years of experience in consulting, IT audit, network engineering, and security. Her focus lies on the technical and organizational aspects of cybersecurity, across various sectors. In her PhD, supervised by Prof. Pavel Laskov, Saskia has developed a core expertise at the intersection of cybersecurity and artificial intelligence, which she is leveraging to drive innovation at Cisco.

Publications:

--Cisco Live Speaker: BRKXAR-1009: Exploring the Paradigm Shift in Security: AI and Post-Quantum's Influence on Zero Trust (2026)

--Schroer, S. L., Apruzzese, G., Human, S., Laskov, P., Anderson, H. S., Bernroider, E. W., & Wang, G. (2025, April). "SoK: On the Offensive Potential of AI." In 2025 IEEE Conference on Secure and Trustworthy Machine Learning (SaTML) (pp. 247-280). IEEE.

--Schroer, S. L., Pajola, L., Castagnaro, A., Apruzzese, G., & Conti, M. (2025). "Exploiting AI for Attacks: On the Interplay Between Adversarial AI and Offensive AI." IEEE Intelligent Systems.

--Schroer, S. L., Canevascini, N., Pekaric, I., Widmer, P., & Laskov, P. (2025, June). "The Dark Side of the Web: Towards Understanding Various Data Sources in Cyber Threat Intelligence." In 2025 IEEE European Symposium on Security and Privacy Workshops (EuroS&PW) (pp. 79-89). IEEE.

--Schroer, S. L., Seideman, J. D., Luo, S., Apruzzese, G., Dietrich, S., & Laskov, P. (2025). "Using a Stack to Find an AI Needle: Topic Modeling for Cyber Threat Intelligence." Digital Threats: Research and Practice.

--Weinz, M., Schroer, S. L., & Apruzzese, G. (2024, September). "'Hey Google, Remind Me to Be Phished' Exploiting the Notifications of the Google (AI) Assistant on Android for Social Engineering Attacks." In 2024 APWG Symposium on Electronic Crime Research (eCrime) (pp. 109-122). IEEE.