What happened
CrowdStrike announced the acquisition of Fiddler AI and Dreadnode at its Fal.Con Europe conference in Barcelona. The deals, valued at approximately $495 million according to the Wall Street Journal, will integrate AI agent observability and pre-deployment security testing into CrowdStrike's Falcon platform. The acquisitions are expected to close in CrowdStrike's fourth quarter of fiscal 2027.
CrowdStrike announced on Monday that it has signed agreements to acquire two artificial intelligence security companies, Fiddler AI and Dreadnode. The announcements were made at the company's Fal.Con Europe conference in Barcelona. While CrowdStrike did not officially disclose the purchase prices, a report in the Wall Street Journal listed the valuation for Fiddler at $400 million and Dreadnode at $95 million.
Fiddler, which specializes in AI agent observability and , will be rebranded as Falcon Observe within CrowdStrike’s Falcon platform. Its technology operates after deployment, evaluating agents before they reach production, tracing decisions and reasoning, flagging drifting behavior or hallucinations, and creating audit-ready records. It supports common agent frameworks such as LangGraph, Amazon Bedrock, and Google ADK.
Dreadnode, which builds automated security testing for AI systems, will integrate its pre-deployment testing capabilities. Its platform runs over 100 known attack methods, including jailbreaks and , against models and agent systems. It ranks identified problems by severity and maps them to security standards like the OWASP Top 10 for large language models and the NIST AI Risk Management Framework.
Alongside the acquisitions, CrowdStrike CEO George Kurtz introduced new Falcon Guardian capabilities, including an 'AI Kill Switch.' This control allows humans to stop any AI agent at runtime, operating outside the model to prevent the agent from rewriting or bypassing the control. Kurtz described this as a shift from traditional security (keeping adversaries out) to safety (allowing AI to run at full speed with human control).
Both products are expected to be generally available in CrowdStrike’s fourth quarter of fiscal 2027. The deals are expected to be paid mostly in cash, with a portion in company stock subject to vesting conditions, and are pending standard regulatory approvals.
Source details: cyberscoop.com
Why it matters
These acquisitions address a critical gap in enterprise security as organizations deploy AI agents that can hold valid credentials yet make erroneous or unsafe decisions. By combining Fiddler's post-deployment observability with Dreadnode's pre-deployment attack simulation, CrowdStrike aims to provide a comprehensive safety layer for AI systems. This move signals a shift in the security industry from traditional perimeter defense to 'safety' controls that allow AI to operate at scale with human oversight, potentially setting new standards for AI governance and risk management in enterprise environments.
The acquisitions target a specific vulnerability in the current AI landscape: the 'black box' nature of agent decision-making. Traditional security tools are designed for predictable code and do not effectively monitor the reasoning processes of AI agents, which can lead to unauthorized actions even with valid credentials.
By integrating Fiddler's observability, CrowdStrike provides enterprises with a 'flight recorder' for AI agents, enabling them to trace exactly what an agent did and why. This is crucial for compliance and audit purposes, as regulators and boards increasingly demand accountability for AI-driven actions.
Dreadnode's integration adds a proactive security layer by simulating attacks before deployment. This helps identify weaknesses in model behavior and prompt handling, reducing the risk of exploitation in production environments.
The introduction of the 'AI Kill Switch' represents a significant conceptual shift in AI safety. By providing a hard stop mechanism that operates outside the model's control, CrowdStrike addresses the concern that AI agents might become uncontrollable or argue against shutdown commands, thereby enabling faster and safer adoption of agentic AI.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
What to watch next
Monitor the general availability of the integrated Falcon Observe and Dreadnode capabilities in CrowdStrike's fiscal 2027 fourth quarter. Watch for regulatory responses to the 'AI Kill Switch' concept and how other security vendors respond to this integrated approach to AI agent safety and observability.
The general availability of the integrated tools in CrowdStrike's fiscal 2027 fourth quarter will be a key milestone. Early adopters' feedback on the effectiveness of the observability and testing features will be important indicators of the product's utility.
Regulatory bodies may scrutinize the 'AI Kill Switch' and the overall safety framework as a potential model for AI governance. This could influence future standards for AI deployment in critical industries.
Competitors in the cybersecurity space may accelerate their own AI security offerings in response to CrowdStrike's move, potentially leading to a broader market shift toward specialized AI safety and observability tools.