Cisco Extends Security Reach to AI Agents

Cisco today at the RSA Conference (RSAC) extended its cybersecurity portfolio to secure artificial intelligence (AI) agents while at the same time employing AI to automate security operations.

At the core of that effort are extensions to the Cisco Duo identity and access management (IAM) platform that make it possible to discover them and apply security policies via a Model Context Protocol (MCP) gateway and the Cisco Secure Access, a security service edge (SSE) platform. That capability will enable cybersecurity teams to assign specific fine-grained permissions that only allow AI agents to access specific resources for a limited amount of time.

At the same time, Cisco unveiled DefenseClaw, an open source secure agent framework that leverages the OpenShell runtime created by NVIDIA to securely automate the deployment of AI agents. The framework includes tools for scanning code created using AI skills tools and MCP integrations along with an AI bill of materials (AI BOM) and…