CrowdStrike unveiled Falcon Guardian at Fal.Con 2026, a runtime security tool designed to monitor and protect AI agents on enterprise endpoints. During a live demo, the product successfully blocked an indirect prompt injection attack targeting Claude Code, intercepting an attempt to exfiltrate AWS credentials before any data left the machine. The system assigned the incident a critical risk score of 90 and identified 12 additional similar attack attempts across the environment, blocking all of them. CrowdStrike currently monitors over 160 million agent instances and has identified more than 1,800 distinct agentic applications, many deployed without formal IT approval.
Source: Read the original article
