cloud strike CEO George Kurtz said Tuesday that cybersecurity companies can now block rogue artificial intelligence agents.
CrowdStrike’s new kill switch, announced at Tuesday’s Fal.Con Europe conference, allows security teams to terminate misbehaving agents, prevent restarts, and cut off network access. It is designed to give humans more power over artificial intelligence models and systems as concerns grow over how to control agents.
“Everybody’s developing AI agents, and they didn’t have an off button, and now they have an off button,” Kurtz told Jim Cramer on CNBC’s “Mad Money.” “This is going to be a really important part of making sure that AI agents run and operate safely.”
What CrowdStrike calls AI Kill Switch is part of the company’s Falcon Guardian platform, which monitors AI agents operating within organizations. The technology is readily available and aims to address AI safety concerns surrounding the rapid deployment of AI agents.
In September, Anthropic CEO Dario Amodei called for a slowdown in AI development citing safety concerns, a position supported by OpenAI CEO Sam Altman. Kurz argued at the time that slowing the development of AI would not eliminate existing security risks and emphasized the need for safety measures when companies deploy autonomous systems.
Kurtz said Tuesday that CrowdStrike’s new kill switch represents a major shift in cybersecurity, which has traditionally focused on protecting businesses from external threats. He argued that as businesses increase their reliance on AI agents, security providers must also ensure that those systems operate securely within their organizations.
“Enabling agents to work safely is a huge market opportunity for CrowdStrike,” Kurtz said. “We have AI capabilities to prevent these breaches from happening externally, and we also have capabilities to make sure our agents are essentially doing what they are told to do.”
The CEO likened the technology to a car’s airbags, arguing that safety measures are still needed even when AI systems are designed to work properly.
“If they step outside the guardrails during execution, we can kill those agents,” he said.
CrowdStrike sees demand for its security tools increasing as enterprises adopt AI. Kurtz pointed to the partnership with CrowdStrike. human To protect Claude Cowork products: He said enterprise customers were asking them to integrate CrowdStrike’s technology before deploying Anthropic’s products.
“Our customers came to Anthropic and said, ‘I need CrowdStrike before I deploy AI,'” Kurtz says. “I think this tells us that customers won’t adopt AI without CrowdStrike.”
CrowdStrike stock closed at a record high of $278.86 on Oct. 6, but has since fallen about 12% as tech stocks struggled in October and November.
