AppSec Aus Melb #23 - Your AI Agent Works for You. Until It Doesn’t.

**Event Confirmation Form!**

➡️ We kindly request you fill out this **event confirmation form** to confirm your attendance for the event and your dietary requirements - [https://forms.gle/2nYqeQmJqz9cvapM7](https://forms.gle/2nYqeQmJqz9cvapM7)

\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-\-

## **What's On? 👀**

**Speaker:** Thomas Roccia

**Title:** Your AI Agent Works for You. Until It Doesn’t.

**Abstract:**

Your AI agent can read your files, call your AP

Is and follow instructions from a page an attacker controls. What could go wrong? In this session, I’ll explore the threats targeting AI agents and demonstrate how Prompt

Intel, NOVA rules and Nova

Hunting help us understand malicious instructions, detect suspicious behavior and investigate agent activity. We’ll also look at the Agent Skill and MCP Scanner to inspect the components we’re trusting. A practical look at what AI threat intelligence means when the software can make decisions and take action.

## **Location** 📍

Atl