AI agents are rapidly being deployed in production, taking real actions with broad access to sensitive data, systems, and cloud resources. But unlike traditional applications, agents don’t simply execute predefined instructions. They reason, adapt, and act based on intent, creating new challenges for security teams
In this session, we’ll explore why traditional security telemetry isn’t enough to understand AI agent actions, and how model input and output telemetry can provide critical visibility into an agent’s reasoning and execution path. We’ll walk through examples of prompt injection and agent abuse, then break down how Wiz is approaching the new security challenges this era brings.
Ponentes: Tal Moriah, Product Marketing Manager, Wiz