Cisco AI Defense & Google ADK: Secure AI Agents

Cisco AI Defense now integrates with Google's Agent Development Kit, known as ADK, to provide runtime protection for AI agents. This means AI Defense security controls can now be applied from local development all the way to the Agent Runtime on the Gemini Enterprise Agent Platform. What's interesting is that agents are powerful because they don't just answer questions; they can use tools and act across multiple steps. This changes the security problem. The real risk is not just a bad answer, but that untrusted content could influence tool behavior or send sensitive data back to the model. Cisco AI Defense addresses this by inspecting and enforcing policy across the full execution path of an agent. This integration aims to add security without disrupting a developer's workflow, allowing them to attach policy enforcement directly to their existing ADK agent structure with just a few lines of code. The bottom line is that this integration helps ensure AI agents are secure as they interact with external systems and handle sensitive data.