CrowdStrike has signed agreements to acquire two artificial intelligence (AI) security startups, Fiddler AI and Dreadnode, in deals valued at a reported $495 million. Fiddler specializes in monitoring and explaining AI agent behavior, while Dreadnode tests AI systems for security weaknesses before they are deployed. The acquisitions are intended to expand CrowdStrike’s protection capabilities for businesses using AI.
The announcement was made on November 3, 2026, during the Fal.Con Europe 2026 cybersecurity event in Barcelona, Spain. CrowdStrike has signed agreements to acquire both companies as it expands its security services for businesses using advanced AI tools. The move comes as more organizations begin using AI agents to handle tasks automatically, creating new security challenges.
The Wall Street Journal reported that CrowdStrike’s two acquisitions are valued at approximately $495 million combined, including around $400 million for Fiddler AI and $95 million for Dreadnode. CrowdStrike did not publicly disclose these individual purchase prices in its acquisition announcements. However, the company confirmed that the payments are expected to be made predominantly in cash, with a portion involving company shares subject to vesting conditions. Both proposed acquisitions are expected to close during CrowdStrike’s fourth quarter of fiscal year 2027, subject to customary closing conditions.
Fiddler AI, founded in 2018 and based in Palo Alto, California, develops technology that helps companies understand how their AI systems work and make decisions. Its software allows businesses to monitor AI agents, track their actions, and identify unusual behavior. This is important because AI agents can now perform complex tasks, access business applications, and make decisions with limited human involvement.
According to CrowdStrike’s Fiddler AI acquisition announcement, the company plans to integrate Fiddler’s technology into its Falcon cybersecurity platform as Falcon Observe. The planned service will help security teams track AI agent actions, understand why certain decisions were made, and identify behavior that may violate company policies. It will also support AI agent evaluation before deployment and ongoing behavioral monitoring. CrowdStrike expects Falcon Observe to become generally available during the fourth quarter of fiscal year 2027.
The second company, Dreadnode, was founded in 2023 and specializes in automated AI security testing. Its technology examines AI models, applications, and agents for security weaknesses before deployment. These tests include prompt injection attacks, where malicious instructions attempt to manipulate an AI system’s behavior, along with attempts to bypass security restrictions, expose sensitive information, or make an AI agent perform unauthorized actions.
In its Dreadnode acquisition announcement, CrowdStrike explains that the technology tests AI systems against security threats such as jailbreak attempts, prompt injection, and attacks targeting autonomous AI agents. The planned integration will help organizations identify security weaknesses before deployment and use those findings to recommend stronger protection policies through Falcon Guardian. CrowdStrike expects its automated AI security testing capabilities to become generally available during the fourth quarter of fiscal year 2027.
The acquisitions are part of CrowdStrike’s wider effort to improve security for AI agents. Unlike traditional software, AI agents can work independently, communicate with other systems, and complete several tasks without constant human supervision. While this can improve productivity, it also creates risks when agents make mistakes, receive harmful instructions, or access information beyond their intended purpose.
A previously reported rogue AI agent incident involving an open-source software project illustrates why AI systems need careful testing, monitoring, and human supervision when they can take actions beyond their intended tasks.
Alongside the acquisition announcements, CrowdStrike introduced an AI Kill Switch in Falcon Guardian, designed to give security teams greater control over AI agents. The feature allows authorized security administrators to interrupt a misbehaving AI agent, prevent it from restarting on selected systems, and restrict its enterprise network access. CrowdStrike announced that the AI Kill Switch was available on November 3, 2026, while additional endpoint-level capabilities are planned for future availability. These controls are intended to help businesses respond more quickly when AI agents behave unexpectedly or create security risks.
The growing use of artificial intelligence in workplaces is increasing demand for stronger security testing, monitoring, and protection tools. Businesses now use AI to handle customer support, analyze information, write software, and automate daily operations. These developments are also part of a broader shift toward AI-powered cybersecurity, as organizations look for faster ways to identify and respond to emerging threats. However, automated systems still require appropriate access restrictions, continuous monitoring, and human oversight to reduce security risks.
CrowdStrike’s planned acquisitions of Fiddler AI and Dreadnode reflect its growing focus on AI agent security. The company aims to combine security testing before deployment, continuous monitoring during operation, and tools that help stop unsafe agent activity. If the acquisitions close and the planned integrations become available, businesses could gain more ways to manage AI security through CrowdStrike’s Falcon platform. However, the effectiveness of the combined tools will depend on their implementation and real-world performance.
