Cisco Launches AI Agent Innovative Solutions, Featuring Assistance for Enterprises Deploying "Zero Trust" Automated Defense Systems

Large corporations are beginning to deploy AI agents, with cybersecurity becoming the top priority before large-scale AI agent usage. Cisco announced an innovative solution at the 2026 RSA Security Conference, focusing on secure deployment of AI agents by integrating identity management and automated security operations. This builds a solid “Zero Trust” defense for enterprises. This article introduces Cisco’s new approach to AI deployment challenges: IAM (Identity and Access Management), AI Defense for AI security, DefenseClaw for automated detection, and Splunk AI for monitoring alerts.

Duo Identity and Access Management (IAM) establishes authentication

Currently, most companies face difficulties in identifying AI agent identities and managing permissions transparently. Cisco extends the “Zero Trust” architecture to AI workloads. Using Duo IAM, companies are required to register AI agents with their responsible owners to ensure traceability. Additionally, Cisco Identity Intelligence automatically detects non-human identities in the environment. Coupled with the Model Context Protocol (MCP) in Secure Access, it limits agent permissions to only what is necessary for specific tasks, preventing security blind spots caused by traditional tools that cannot understand the context of agent requests.

AI Defense: Providing development teams with model security testing

To address the risks of malicious prompts or tampering with AI agents, Cisco introduces “AI Defense,” a self-service tool that allows developers and security teams to perform stress tests on models before deployment. The toolkit supports dynamic red team exercises to verify application resistance against malicious outputs and generate compliance reports. Cisco also releases the Agent Runtime SDK and a security ranking for large language models (LLMs), linking model performance with risk indicators. This helps organizations objectively assess the security of different AI models and embed security strategies into workflows during the development phase.

DefenseClaw promotes open-source security frameworks and automated sandbox environment integration

To minimize the impact of security reviews on development progress, Cisco launches “DefenseClaw,” an open-source security agent framework for automating security management and asset scanning. The framework integrates tools like skill scanning and AI Bill of Materials (AI BoM). Plans include integration with NVIDIA OpenShell, utilizing sandbox isolation technology to eliminate manual security checks. Through open-source collaboration and automation, enterprises can accelerate AI agent deployment while maintaining the integrity of Zero Trust, ensuring all components are verified and isolated.

Splunk AI embeds new AI features for automatic alert detection

Facing automated attack challenges, Splunk AI has embedded several new AI capabilities into the Security Operations Center (SOC) framework, shifting from passive defense to active interception. The new intelligent SOC extension includes detection building, malware reverse analysis, and automated response agents, automating tasks like alert classification and threat investigation. Through exposure analysis and joint search functions, analysts can correlate data across environments and receive real-time risk scores. This enables security teams to handle complex threat workloads at machine speed, enhancing overall resilience.

This article about Cisco’s innovative AI agent solutions, focusing on helping enterprises deploy “Zero Trust” automated defense systems, first appeared on Chain News ABMedia.