CrowdStrike has introduced a security tool that allows businesses to immediately terminate artificial intelligence agents that behave unexpectedly, as companies seek stronger controls over autonomous systems being deployed across their operations.
Chief Executive George Kurtz announced the AI Kill Switch on Tuesday at the company’s Fal.Con Europe conference. The tool enables security teams to stop misbehaving agents, prevent them from restarting, and disconnect them from corporate networks.
The feature is part of CrowdStrike’s Falcon Guardian platform, which monitors AI agents operating within an organization. Available immediately, it is designed to give businesses direct intervention capabilities when autonomous systems act outside their authorized instructions or established security boundaries.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
“Everyone is building AI agents, but they never had the off button, and now they do,” Kurtz told CNBC’s Jim Cramer on Mad Money. “It’s going to be an incredibly important part of making sure that AI agents run and operate safely.”
The launch addresses a growing security challenge as companies move beyond conventional chatbots towards AI agents capable of executing tasks, accessing business applications and interacting with corporate systems. Unlike systems that only generate responses, agents can take actions on behalf of users, creating additional risks when they operate with excessive permissions, encounter malicious instructions or behave in ways their developers did not anticipate.
CrowdStrike’s approach gives security teams a mechanism to interrupt that activity rather than relying solely on preventive controls.
Kurtz said the introduction of the kill switch represents a change in the responsibilities of cybersecurity providers. Traditional cybersecurity has largely focused on protecting organizations from external attacks, including malware, data theft, and unauthorized access. The growing use of autonomous agents creates another security requirement of ensuring that systems operating inside corporate networks remain within approved boundaries.
CrowdStrike argues that businesses need controls capable of monitoring and stopping AI agents while they are running, particularly as those systems gain access to sensitive information and operational tools.
“Enabling agents to work safely is a massive market opportunity for CrowdStrike,” Kurtz said. “We have AI capabilities to prevent these breaches from happening on the outside, and we also have the capabilities to make sure that agents are basically doing what they are instructed to do.”
He compared the new safeguard to airbags in vehicles, saying protection mechanisms remain necessary even when systems are designed to operate correctly.
“If they go outside of those guardrails at runtime, we can kill those agents,” he said.
Restricting an agent before deployment does not necessarily guarantee that it will behave as intended during operation. Agents can encounter unexpected inputs, interact with connected applications in unintended ways, or be manipulated through malicious instructions. A shutdown mechanism provides an additional layer of response when preventive measures fail.
However, the kill switch is not a substitute for broader AI governance. Businesses still need to define what agents are authorized to access, monitor their activities, manage permissions, and establish procedures for investigating incidents. Terminating an agent can stop its further activity, but it does not automatically reverse actions already completed or recover information that may have been exposed.
CrowdStrike is positioning the product within a broader commercial opportunity as businesses increase their use of AI tools. Kurtz cited the company’s partnership with Anthropic to secure its Claude Cowork product, saying enterprise customers had asked for CrowdStrike’s technology to be integrated before adopting Anthropic’s offerings.
“Our customers went to Anthropic and said, ‘We need CrowdStrike before we can deploy AI,’” Kurtz said. “I think it says volumes that customers would not roll out AI without CrowdStrike.”
The comments indicate that security requirements are becoming part of enterprise AI procurement decisions. Companies evaluating autonomous systems must consider not only their productivity benefits but also the risks associated with granting software agents access to internal data, applications and workflows.
The launch also comes amid wider debate over the pace of AI development and the safeguards needed to manage capable systems. In September, Anthropic Chief Executive Dario Amodei called for slowing AI development over safety concerns, a position backed by OpenAI CEO Sam Altman. Kurtz argued at the time that slowing development would not remove security risks posed by systems already in use, making practical safeguards necessary.
CrowdStrike shares closed at a record $278.86 on October 6 but have since fallen about 12% as technology stocks came under pressure in October and early November. The new product adds another potential growth area for the cybersecurity company as it seeks to expand its role in protecting businesses adopting AI.