Presented by
About this talk
The Model Context Protocol (MCP) extends LLMs by integrating external tools but introduces serious security risks like Tool Poisoning and Rug Pull attacks. This session will explore these vulnerabilities and present layered defenses, including cryptographic tool verification, immutable definitions, and fine-grained policy-based access control, highlighting MCP as a critical AppSec challenge. Join Gianpietro Cutolo at RSAC 2026 to learn more.