Las Vegas – October 31, 2026 -- CrowdStrike's Falcon Guardian has been deployed at hundreds of organizations since its September 1, 2026 launch, and early users report finding over 260 distinct AI services at a single organization and confidential documents being shared with public AI chatbots.
Customers span financial services, healthcare, manufacturing, retail, energy and the public sector. They are responding to the shift from AI that generates content to agents that reason, decide and act autonomously. Employees run AI assistants in browsers and desktop apps, developers run coding agents that read files and execute commands, and business teams connect agents to enterprise data through Model Context Protocol (MCP) servers.
Early deployments uncover hundreds of unrecorded AI interactions
Visibility was the first concern for most customers. Security leaders knew AI use was widespread but could not see which agents and tools were running, what data entered prompts, or which tools were sanctioned versus shadow AI. Early deployments turned up hundreds of AI interactions that existing controls had not recorded.
Customers that already run the Falcon sensor and hold a Falcon Guardian license get visibility within minutes. Several deployed it live in production during workshops at Fal.Con 2026 in Las Vegas and saw results immediately.
Endpoint telemetry links agent prompts to system actions
Falcon Guardian works through the existing Falcon sensor and console, plus the Falcon browser extension. Many customers cited endpoint coverage as the deciding factor. According to the source, tools that inspect network traffic or integrate with each agent through hooks cannot connect an agent's AI activity to the system actions it triggers. Falcon Guardian combines AI telemetry with endpoint telemetry to show what an agent was asked to do and what it did. One CISO described CrowdStrike as "uniquely positioned" because it is already on the endpoint.
Evaluation finds two internal agents unknown to security operations
In one customer evaluation, Falcon Guardian found two internally built agents that the security operations team did not know were running. Another organization used it to monitor MCP activity and detect risky data and instructions passing between agents and MCP servers.
CrowdStrike's Falcon Adversary OverWatch threat hunting team also observed an endpoint agent try to use public data-sharing websites to share corporate IP files to meet its objective. Security teams can set which supported agent types may run, enforce runtime guardrails, investigate activity in the Falcon console and respond to unexpected behavior.
Customer tests record 99% prompt injection detection accuracy
Detection draws on CrowdStrike's prompt injection taxonomy, which covers more than 200 techniques. A financial services firm measured 99% detection accuracy on prompt injection. A global telecommunications company measured average block rates above 95% across malicious prompt categories, with multilingual coverage above 98%.
Policies can detect, redact or block personal and confidential information in supported AI interactions. Controls can be applied by user, application and other attributes.
Governance requirements drive rollout of AI assistants to 20,000 employees
One organization made Falcon Guardian a requirement before deploying an AI assistant to more than 20,000 employees. Other customers needed to show how AI use is governed under HIPAA, DORA, NIS2 and the EU AI Act.
Falcon Guardian can be added to existing CrowdStrike deployments. Its findings flow into CrowdStrike Falcon Next-Gen SIEM, so AI activity can be investigated alongside endpoint, identity, SaaS and cloud data.