Generative AI and autonomous AI agents are rapidly becoming part of modern cloud-native applications. However, as organizations integrate Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), APIs and Model Context Protocol (MCP) servers into their environments, they introduce a new and evolving attack surface.

Traditional application and network security controls are not always sufficient to address threats targeting AI applications.

This session explores the emerging AI security landscape and demonstrates how organizations can identify, understand and mitigate threats against LLM-powered applications and AI agents.

We will examine practical attack scenarios, including prompt injection, jailbreaks, sensitive information disclosure, system prompt leakage, insecure tool execution and excessive agent permissions. We will also explore the role of AI firewalls, runtime inspection, network segmentation and policy enforcement in protecting AI workloads.

Attendees will gain a practical understanding of how to build a layered security strategy for AI-powered applications without compromising innovation and productivity.