Cisco Addresses Security Risks of AI Agents at RSA Conference 2026

• •85% of enterprises have experimented with AI agents, but only 5% have deployed them.
• •Cisco's new security capabilities focus on identity management and Zero Trust for AI agents.
• •Real-time policy enforcement is essential to govern the actions of autonomous AI agents.

Cisco has launched new security capabilities to address the risks posed by AI agents, which are increasingly being integrated into customer experience workflows. While 85% of enterprises have experimented with AI agents, only 5% have deployed them at scale due to concerns over control and accountability. Jeff Schultz from Cisco highlighted that unlike chatbots, which primarily posed risks through their responses, AI agents can autonomously perform actions, raising significant security challenges. Cisco's new offerings include enhanced identity management, Zero Trust access controls, and real-time policy enforcement to govern agent behavior. The company aims to ensure that AI agents are treated similarly to human workers in terms of governance and security. Cisco's approach emphasizes the need for organizations to understand what AI agents are doing in real-time to mitigate potential blind spots. The announcements were made during the RSA Conference in San Francisco, where Cisco is actively addressing the evolving landscape of AI in cybersecurity.