CrowdStrike invests $495M in AI agent security with Fiddler, Dreadnode deals

• CrowdStrike to acquire Fiddler AI and Dreadnode for $495 million, boosting protection for autonomous AI agents.
• Falcon platform will introduce agent observability, automated AI testing, and an AI Kill Switch for advanced risk mitigation.

On November 4, 2026 (UTC), The Wall Street Journal reported that CrowdStrike will acquire AI security startups Fiddler AI and Dreadnode in a combined $495 million deal, announced during its Fal.Con Europe conference. This move signals CrowdStrike’s largest direct investment to date in shoring up cybersecurity defenses for the rapidly expanding field of autonomous AI agents.

Fiddler AI powers agent observability, letting security teams retrace and explain autonomous agent decisions—an increasingly urgent capability as organizations embed agents into core IT and finance systems. These features will be deployed under Falcon Observe to give companies transparency in high-stakes, automated environments, such as algorithmic trading or healthcare diagnostics.

Dreadnode’s technology automates pre-deployment vulnerability testing specifically for AI agents. This allows Falcon Guardian to systematically evaluate an agent’s code and behavior for exploitable flaws before deployment, a critical need as AI systems outpace traditional manual review timelines.

CrowdStrike also introduced an AI Kill Switch in Falcon Guardian, enabling administrators to immediately shut down rogue or malfunctioning agents, block their restart on selected endpoints, and halt their network connections. The switch is already live at the AI gateway level, with plans for deeper endpoint integration. This provides organizations with real-time intervention tools, illustrated by recent incidents where malfunctioning agents corrupted financial transactions or exposed customer data before IT teams could respond.

These new features target the unique and rapidly evolving risks presented by autonomous AI agents. Industry research cited by OWASP and BCG notes that only 23% of organizations monitor AI agent activity, leaving systems exposed when legitimate agents act outside prescribed parameters. Since traditional endpoint security cannot always detect or control these emergent behaviors, CrowdStrike’s investment reflects growing demand for pre-deployment vetting, real-time monitoring, and live containment measures.

CrowdStrike expects to close the acquisitions and deliver integrated features in the fourth quarter of its fiscal 2027. This announcement arrives amid McKinsey projections that cybersecurity spending on AI protection will climb from 4% to 15% within three years, heralding greater competition and innovation among top industry providers.