CrowdStrike (NASDAQ: CRWD) announced new capabilities for its Falcon platform that expand AI agent discovery and runtime protection across endpoints, software-as-a-service platforms, browsers, and cloud environments.

The cybersecurity company said its sensors detect more than 1,800 distinct AI applications running on enterprise devices, representing nearly 160 million unique application instances across its customer base. The new features aim to provide visibility and control over AI behavior as autonomous systems gain system-level privileges.

"AI agents are fundamentally changing how technology operates and how it must be secured," said Michael Sentonas, president of CrowdStrike. "Security built for static applications can't keep up with autonomous systems."

The endpoint security enhancements include EDR AI Runtime Protection, which captures commands, scripts, file activity, and network connections of all applications running on endpoints. Shadow AI Discovery for Endpoint automatically identifies AI applications, agents, and development tools across endpoints. AIDR for Endpoint extends prompt-layer protection to desktop AI applications including ChatGPT, Gemini, Claude, and Microsoft Copilot.

The platform expansion covers additional environments through CrowdStrike's acquisition of Seraphic Security, which extends runtime protection to browsers. New cloud capabilities include Shadow AI Discovery for Cloud and AIDR for Cloud, which secures AI workloads in containerized environments.

Additional features include Shadow SaaS and AI Agent Discovery for platforms like Microsoft Copilot, Salesforce Agentforce, and ChatGPT Enterprise, and AIDR for Copilot Studio Agents that monitors prompts and agent behavior in real time.

The announcement was made at RSA 2026 conference. The press release includes a disclaimer that some referenced services or features are still in development and subject to change.