Abstract
AI systems (LLMs, Agents, MCP servers) are rapidly becoming central to modern applications, but with this evolution comes a new class of security risks. Our hands-on activity workshop, AI Security Playground, is designed to help developers learn about security challenges unique to AI and GenAI environments. In this interactive workshop, participants will get their hands dirty with real world attacks against AI from beginner to advanced including sophisticated exploits against AI agents, model context protocol (MCP) server misconfigurations and classic to advanced prompt injections.
This will be a hands-on-session, powered by our open-sourced, gamified training platform, attendees will explore vulnerabilities step-by-step, learn how attackers think, and master AI specific defensive techniques. Whether you’re an AI practitioner, security enthusiast, or developer, this workshop promises to level up your understanding through practical challenges, insightful explanations, and immediate feedback as we offer an AI powered judge on the platform who would give you feedback once you solve each challenge.
This hands-on Arsenal Lab session was also scheduled on Friday, April 24 | 1:00pm-1:30pm (Arsenal Lab Station 2, Business Hall).