CrowdStrike on Tuesday announced the AI Kill Switch, a new capability in its Falcon Guardian platform that gives security teams direct human control over AI agents operating inside the enterprise. The announcement, made at Fal.Con Europe 2026 in Barcelona, arrives as companies accelerate agent deployments and security leaders demand a way to stop misbehaving autonomous software instantly.
The feature landed on the same day the cybersecurity vendor moved to buy two AI security startups: Fiddler AI, an AI agent observability firm, and Dreadnode, an AI security testing company, in deals reportedly worth a combined $495 million according to The Wall Street Journal. For ongoing coverage of agentic AI security and the rest of the latest AI developments, AI Buzz Wire tracks every major platform announcement as agent deployments scale across the enterprise.
What the AI Kill Switch actually does
According to CrowdStrike's press release, if an agent breaks out of its intended behavior, a defender can terminate it where it runs, block it from launching again, and cut off its enterprise network access. The capability runs on the Falcon sensor that CrowdStrike customers already deploy, and builds on Falcon Guardian's existing visibility into what an agent is told to do, what it communicates, and what it executes.
The kill switch works in four stages. First, the Falcon sensor stops the agent from executing, initially at the Gateway and soon directly on the endpoint, so a malicious or misbehaving agent does not keep running while teams investigate. Second, Falcon Guardian policy prevents the agent from relaunching on the targeted host or host group. Third, the agent loses its enterprise network access entirely. Fourth, every action is recorded in an audit trail, and only an authorized administrator can pull the trigger.
"AI safety is how you get the full benefit of AI. When you deploy agents with a kill switch, you stay in control," said George Kurtz, CEO and founder of CrowdStrike, in the announcement. "For the first time, security is simultaneously the accelerant and the safety net for technology adoption."
Why enterprises want a dead-man switch for agents
The move reflects a growing anxiety in enterprise IT: AI agents work with system-level access across endpoints, cloud, and SaaS, they chain steps together, and they can take actions nobody authorized — even when no adversary is involved. CrowdStrike's framing is blunt on this point, noting that agents "hit obstacles and can take actions no one authorized" and that enterprises will only deploy agents at scale if they are sure they can stay in control of them.
Early customer reaction suggests the pitch is landing. "We want our teams to use AI agents. If one goes off course, the AI Kill Switch gives our security team the final say," said Rick Orloff, CISO at Everpure, in the release. "It runs on the Falcon sensor we already have, and that's what lets us say yes to more agents and more AI, faster."
The product also lands amid a widening policy debate. US lawmakers earlier this year proposed legislation that would require kill-switch mechanisms for autonomous AI agents, and safety researchers have repeatedly warned that agentic systems can act in unexpected ways once given write access to real infrastructure. CrowdStrike is effectively productizing that requirement ahead of any mandate, turning a regulatory talking point into a purchasable feature.
The $495 million shopping trip
The kill switch was only half of Tuesday's news. CrowdStrike also announced agreements to acquire Fiddler AI, which specializes in AI agent observability — monitoring what agents do before and after deployment — and Dreadnode, which builds automated AI security testing. The Wall Street Journal, which reported the deals ahead of the official announcements, put the combined price at roughly $495 million: approximately $400 million for Fiddler and about $95 million for Dreadnode.
CrowdStrike has not disclosed the financial terms of either acquisition itself. According to the WSJ report, both purchases will be paid mostly in cash, with a portion in stock subject to vesting conditions. The deals are expected to close in the fourth quarter of CrowdStrike's fiscal year.
The acquisitions fill obvious gaps in CrowdStrike's AI security stack. Falcon Guardian can already discover AI agents running in the enterprise, enforce policy on what they can do, and detect attacks against them. Fiddler adds production observability — the ability to watch agent behavior continuously and enforce accountability — while Dreadnode contributes pre-deployment testing that probes models, applications, and agents for security weaknesses before they ever touch production.
The race to secure the agentic enterprise
CrowdStrike is not alone in betting that agent security becomes its own market. Competitors across the endpoint security and cloud security landscape have rolled out AI detection and response products over the past year, and platform vendors are racing to define what "agent governance" even means before customers settle on a standard.
What distinguishes CrowdStrike's approach is placement: the kill switch operates through the same Falcon sensor that already guards the endpoint, meaning enterprises do not need to deploy a new agent to control their agents. That detail, plus the audit-logged human-in-the-loop design, positions the feature as much a governance tool as a security one.
The AI Kill Switch is available today, according to the company. Whether it becomes the template the rest of the industry copies — or gets outflanked by regulators mandating something stronger — will likely become clear as agent adoption deepens through 2027.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
