(Virtual) When An Agent Goes Rogue: Securing Agentic AI Workshop Intro
AI agents are no longer limited to generating text or recommendations. Modern LLM powered autonomous agents now execute real world actions such as moving money, invoking privileged APIs, modifying infrastructure, and operating across production systems.
This AI focused virtual session explores how agentic AI systems behave at execution time, where risk actually emerges and where traditional AI guardrails fall short.
In this workshop, we will examine how autonomous AI agents execute actions across tools, APIs, and workflows. We will break down common failure modes in production agentic systems including tool misuse, cascading actions, and unintended execution. We will also explain why policies, prompts, and post execution audits cannot control AI at the moment an action is taken, and how execution time authorization introduces a new control point for AI systems.
This session is designed for AI builders, platform teams, and engineers working with LLM based agents, autonomous AI workflows, AI infrastructure and tooling, and production AI systems that interact with real world resources.
What you will learn
You will leave with a clear mental model for how agentic AI executes actions in production environments. You will learn practical frameworks for controlling high risk AI behavior at execution time, understand how execution time controls complement existing AI guardrails, and see architectural patterns for safely deploying autonomous agents.
Speakers
Ben Zuiker is the Co Founder and CEO of QSVA. He is a multi time founder and cryptographic systems inventor focused on execution time authorization and verification for AI systems. Ben previously co founded Zmt Labs, a cryptography research lab dedicated to authentication and authorization systems resilient to AI driven and post quantum threats.
Tom Gilheany is the Co Founder and CTO of QSVA. He is a security and AI infrastructure leader with more than 30 years of experience building and securing enterprise systems across financial services and regulated industries. Tom previously served as Vice President of Wells Fargo Cybersecurity Innovation and now leads execution time security architecture for agentic systems at QSVA.
Prof PK Prasanna Kumar is the Chief AI Officer at QSVA. He is a serial entrepreneur, senior executive, and applied AI expert across healthcare, finance, and large scale AI platforms. He is a bestselling author and adjunct professor specializing in applied AI deployment, governance, and datacenter scale AI infrastructure.
Format
This is a 90 minute live virtual session featuring interactive walkthroughs, architectural discussion, and real world examples from production AI systems.
Hosted by the QSVA leadership team, this session introduces execution time security controls for agentic AI and provides a preview of hands on workshops where teams design and stress test controls for their own autonomous systems.
#ai