Securing the AI Revolution - [...]
summary
Argues MCP is functionally an API but crucially different: agentic consumers, stateful streaming, and a dynamic contract that enables real-world interactions and new attack vectors like tool poisoning and rugpull updates. Recommends an MCP gateway/broker that reconciles streaming variants, enforces gated discovery of tools, blocks dangerous tool-call sequences, mediates tokens for zero trust, applies token-aware metering and identity-based policies, and captures telemetry; demo shows a control plane detecting and approving a contract drift before a new tool becomes available.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:12MCP in AI landscape
- 3:21MCP vs API
- 5:09MCP threats
- 8:04Tool poisoning risks
- 9:49MCP governance security
- 13:50MCP demo setup
- 14:55API to MCP conversion
- 16:00Secure MCP contract changes
- 17:30Agent authorization flows
- 19:18MCP observability
- 19:45MCP governance demo
- 22:58Conclusion
- 24:03Q&A
related talks
17mBuilding Production-Ready MCP Servers: From AI Agents to Secure Enterprise APIs.
50mMCP at Scale: Governing AI Agent APIs When Autonomous Systems Are Your Consume…
45mAPIs are the best protection for your AI strategy -
29mMCP Security: Keep Your AI Agents from Spilling the Tea
15mWhy Security Matters: The Risks of Agentic AI and How to Mitigate Them
22mSAFE-MCP: A Security Framework for AI+MCP (Model Context Protocol)
45mSecuring MCP Servers with Spring AI
34mBalancing Security and Innovation in Multi-Agent Systems: An Open Source Approach