MCP Under Fire: Threats and Countermeasures for AI Tool Integrations

No ratings

Presented at BSidesCharm 2026 by

Jonathan Butler

Paul Navarro

The Model Context Protocol (MCP) is gaining traction, but what are its security risks? This talk provides a technical deep dive into MCP. We'll explore the MCP core architecture, common server challenges, and MCP vulnerabilities. We will then shift to defense, covering official security best practices and demoing open-source tools to secure your MCP ecosystem. Whether you're using or building AI agents, deploying MCP servers, or securing AI infrastructure, this talk provides the security insights you need to safely harness MCP's potential.