CrowdStrike adds kill switch to stop rogue AI agents

The news: CrowdStrike on Nov. 3 announced the AI Kill Switch in Falcon Guardian, a feature it says gives “security teams the ultimate human control over AI agents.” Falcon Guardian is the part of CrowdStrike’s Falcon security platform that, according to the company, “discovers the AI agents running in the enterprise, enforces policy on what they can do, and detects attacks against them.” The announcement came from Austin, Texas, and CrowdStrike’s Fal.Con Europe 2026 conference in Barcelona.

What it does: If an agent “breaks out of its intended behavior,” CrowdStrike says, a defender can “terminate it where it runs, block it from launching again, and cut off its enterprise network access.” The company lists four parts. The Falcon sensor, CrowdStrike’s software that runs on customers’ machines, “stops the agent from executing, first at the Gateway, and soon on the endpoint.” Falcon Guardian policy “keeps the agent from relaunching on the targeted host or host group.” The agent “loses its enterprise network access.” And “an authorized administrator triggers the action, and every action is recorded in the audit trail.”

What it can’t do yet: The timing language matters. CrowdStrike says the termination works “first at the Gateway, and soon on the endpoint.” The release does not say when “soon” is. It also says “The AI Kill Switch is available today,” and elsewhere that the feature “starts on the endpoint, where the Falcon sensor already operates.” The release does not reconcile those lines, and its forward-looking statement notes that “any unreleased services or features referenced here are still in development and subject to change.”

Why CrowdStrike says it’s needed: The company argues that AI agents “work with system-level access, across endpoints, cloud, and SaaS,” and that they “chain steps, hit obstacles, and can take actions no one authorized, even when no adversary is involved.” Its view: “Enterprises will only deploy agents at scale if they are sure they can stay in control of them.”

Who’s talking: “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. Rick Orloff, CISO at Everpure, said in the release, “If one goes off course, the AI Kill Switch gives our security team the final say,” adding that “it runs on the Falcon sensor we already have.”

Who it’s for: Security teams at companies already running CrowdStrike’s Falcon sensor that want to let employees use AI agents without losing the ability to stop one.

Our read: The pitch here is less about detecting attacks than about giving a person a reliable off switch for software that acts on its own. That framing fits a real worry: CrowdStrike’s own description of agents taking “actions no one authorized, even when no adversary is involved” describes a failure mode that classic malware tools weren’t built around. Bundling termination, a relaunch block and network cutoff into one logged, human-triggered action is a sensible design, and building on a sensor customers already run lowers the bar to adoption. But buyers should read the fine print. The strongest version of the promise, stopping an agent directly on the machine where it runs, is described as coming “soon,” with no date, while the gateway enforcement comes first. The release offers one customer quote and no performance data, so how well it catches and stops misbehaving agents in practice is still CrowdStrike’s claim to prove.

What’s next: CrowdStrike says the AI Kill Switch “is available today” and points readers to a companion release on “new Falcon Guardian innovations.” The release lists no pricing. Watch for endpoint termination to arrive and for a date on what “soon” means.

Sources: CrowdStrike: CrowdStrike Introduces the AI Kill Switch for AI Safety (Nov. 3, 2026)