As enterprises race to deploy autonomous AI agents across their operations, Cisco has stepped in with a comprehensive security framework designed to protect this emerging agentic workforce. Unveiled at RSA Conference 2026 in San Francisco, the announcement centers on DefenseClaw, a new open-source framework, alongside AI Defense: Explorer Edition and a suite of Splunk-powered security automation tools.
DefenseClaw: Open-Source Agent Security
DefenseClaw is an open-source secure agent framework that automates security inventory and hardening for enterprise AI agents. The framework integrates with NVIDIA OpenShell as a sandboxing environment, eliminating the manual security configuration steps that currently slow down agent deployment.
By open-sourcing the framework, Cisco is betting that community-driven development will produce more robust security tooling than proprietary alternatives — a significant move from a company historically known for closed-source enterprise products. The framework addresses a growing concern: as AI agents gain autonomy to execute tasks, browse the web, write code, and interact with enterprise systems, the attack surface expands dramatically.
AI Defense: Explorer Edition
Alongside DefenseClaw, Cisco introduced AI Defense: Explorer Edition, which democratizes AI safety testing by providing developers with self-serve tools to probe model and application resilience against attacks. The platform allows teams to embed robust guardrails into agents before they reach production, addressing the gap between rapid agent development and security validation.
The tool targets a critical pain point: most enterprise AI agent deployments currently lack standardized security testing, leaving organizations exposed to prompt injection, data exfiltration through agent tool use, and unauthorized action execution.
Splunk AI Security Automation
Cisco also announced new Splunk AI innovations that transform security operations by automating response workflows. The integration enables security teams to respond to threats at machine speed, a necessity as AI-powered attacks become more sophisticated and frequent.
Staggered Rollout
The product suite follows a phased release schedule. Detection Studio and the Malware Threat Reversing Agent are generally available now. Exposure Analytics, SOP Agent, and Federated Search are expected to launch in April and May 2026. The Automation Builder Agent, Triage Agent, Detection Builder Agent, and Guided Response Agent are targeting June 2026, with some entering prerelease testing.
Why Agent Security Matters Now
The timing is significant. With NVIDIA's OpenClaw, Microsoft's AutoGen, and a growing ecosystem of agent frameworks making it easier than ever to build autonomous AI systems, security has emerged as the critical bottleneck to enterprise adoption. A recent Cisco report found that the majority of enterprises deploying AI agents lack adequate security controls, and that agentic systems introduce attack vectors that traditional cybersecurity tools were never designed to handle.
Cisco's bet is that whoever establishes the security standard for the agentic era will capture a market that barely existed a year ago but is now central to enterprise AI strategy. For organizations still deciding where to deploy agents, this guide on AI agents in the enterprise covers where to start and what pitfalls to avoid.
