Fortune 500 company discovers 18,000 unapproved AI agents in systems.

A Fortune 500 company discovered 18,000 active AI agents on its systems when it activated agent discovery in August, despite having only approved 300. This revelation was highlighted by Michael Sentonas, President of CrowdStrike, during his keynote speech at Fal.Con 2026, where he introduced Falcon Guardian, a new security feature.

The timing of this discovery underscores the growing complexity of managing AI technologies within organizations. As businesses increasingly rely on AI tools, the risk of unmonitored or unauthorized agents—often referred to as shadow AI—has escalated. The surge in AI adoption has been driven by a burgeoning underground economy of applications and skills that operate outside traditional IT visibility. This trend raises significant concerns about data security and compliance, prompting companies to reassess their AI governance strategies.

Among the identified agents were Claude Code, OpenAI Codex, Cursor, and Kiro, but CrowdStrike did not disclose how many of the 17,700 unidentified agents could be classified as shadow AI. The company noted that the growth of these hidden agents is a direct result of the rapid expansion of AI tools and platforms that operate without official oversight. The situation emphasizes the need for organizations to enhance their monitoring capabilities to safeguard sensitive data and maintain control over their technological environment.

Looking ahead, companies must prioritize understanding the full scope of AI agents in use. Falcon Guardian, part of CrowdStrike’s Falcon Flex licensing model, allows existing customers to activate this discovery capability immediately. As businesses navigate the challenges posed by an increasing number of AI agents, implementing robust security measures will be crucial for ensuring operational integrity and protecting against potential threats.