CrowdStrike launches tool to stop out-of-control AI agents

CrowdStrike shares fell approximately 12% from a record high of $278.86 set on October 6. This decline occurred as the company announced a new tool designed to manage the growing risk of autonomous software.

The tool, named the AI Kill Switch, allows security teams to terminate AI agents that behave unexpectedly. It blocks these systems from restarting and severs their network access. The feature is part of Falcon Guardian, a platform used to monitor AI agents within organizations. The tool is available to customers immediately.

George Kurtz, the chief executive officer, introduced the feature on Tuesday at the Fal.Con Europe conference. He told CNBC's Jim Cramer on the show Mad Money that the industry lacked a simple way to stop these systems. "Everyone is building AI agents, but they never had the off button, and now they do," Kurtz said. He described the tool as an incredibly important part of ensuring safe operation.

The announcement coincides with a wider debate about controlling autonomous systems. In September, Dario Amodei, CEO of Anthropic, urged the industry to slow development due to safety concerns. Sam Altman, CEO of OpenAI, supported this stance. Kurtz argued that slowing development would not eliminate existing security risks. He maintained that safeguards are necessary as companies deploy autonomous systems.

Kurtz stated that the kill switch represents a shift in cybersecurity. The industry has historically focused on defending against outside attackers. As businesses rely more on AI agents, providers must also ensure these internal systems behave correctly. "Enabling agents to work safely is a massive market opportunity for CrowdStrike," he said. The company uses the technology to prevent external breaches and to verify that agents follow their instructions.

Kurtz compared the safeguards to airbags in cars. He noted that protections remain necessary even when systems are built to function correctly. "If they go outside of those guardrails at runtime, we can kill those agents," he said. He cited demand from enterprise customers, specifically referencing CrowdStrike's partnership with Anthropic to protect the Claude Cowork product.

Kurtz said customers told Anthropic they required CrowdStrike before deploying AI. "I think it says volumes that customers would not roll out AI without CrowdStrike," he said. The company's stock price movement in recent weeks reflects the current struggle of technology stocks through October and into November.