Vandana Verma Sehgal presented “OWASP MCP Top 10: When AI Agents Go Rogue, Securing the Model Context Protocol” at OWASP Global AppSec EU 2026 in Vienna.
The session examined how attackers can manipulate agent context, poison tool outputs, inject malicious instructions, and exploit the new trust boundaries introduced by MCP-enabled systems. It also covered defensive patterns for securing agentic AI architectures.
Official session
https://owaspglobalappseceuvienna20.sched.com/event/2Kwsu/owasp-mcp-top-10-when-ai-agents-go-rogue-securing-the-model-context-protocol