CrowdStrike CEO George Kurs ( George Kurtz ) stated on Tuesday that the cybersecurity company is now capable of shutting down out-of-control artificial intelligence agents ( AI Agent ).

On Tuesday, at the Fal.Con conference in Europe, a new “emergency stop switch” (Kill Switch) feature was announced. This feature allows security teams to terminate AI intelligents with abnormal behavior, prevent them from restarting, and cut off their network access rights.

This technology is aimed at enhancing human control over artificial intelligence models and systems. With the increasingly widespread use of AI intelligents, how to effectively control these autonomous systems is attracting more and more attention.

Kurtz said in an interview on Tuesday: "Everyone is developing AI intelligents, but in the past they never had a shutdown button, whereas now they do. This will become an extremely important aspect in ensuring the safe operation and functioning of AI intelligents."

CrowdStrike named this feature AI Kill Switch, which is part of the company's Falcon Guardian platform. This platform is responsible for monitoring the AI intelligents that operate within enterprises.

This technology has now been officially launched, aiming to address the artificial intelligence security issues brought about by the rapid adoption of AI intelligents by enterprises.

In September this year, Anthropic CEO Dario Amodei expressed concerns regarding security and called for a slowdown in the development of artificial intelligence, a stance that was also supported by OpenAI CEO Sam Altman.

At that time, Kurz retorted that slowing down the development of artificial intelligence would not eliminate existing security risks. He argued that this actually highlights the necessity for companies to establish security protection mechanisms when deploying autonomous systems.

Kurtz stated on Tuesday that the emergency stop feature introduced by CrowdStrike represents a major shift in the field of network security. Traditional network security has mainly focused on protecting companies from external threats, but as companies become increasingly dependent on AI intelligents, network security service providers must also ensure that these systems operate securely within the companies.

Kurtz stated: "Ensuring that the AI intelligent entity can operate safely has brought tremendous market opportunities for CrowdStrike. We possess the technology to utilize artificial intelligence capabilities to prevent external cyberattacks, and we also have the ability to ensure that the AI intelligent entity acts in accordance with instructions."

This CEO compares this technology to the airbag in a car, believing that even though artificial intelligence systems are designed with the requirements for correct operation in mind, it is still necessary to establish safety protection mechanisms.

He said, "If these agents break through security boundaries during actual operation, we can terminate them directly."

As enterprises continue to adopt artificial intelligence, the security tools of CrowdStrike are attracting increasing attention from more and more customers.

Kurtz specifically mentioned the collaboration between CrowdStrike and Anthropic. The two parties worked together to provide security protection for Claude Cowork products of Anthropic. He stated that before deploying the products of Anthropic, corporate clients actively requested the integration of CrowdStrike technology.

Kurtz said, "When our clients directly refer to Anthropic, it means: 'Before deploying AI, we need to deploy CrowdStrike first.' I believe this fully demonstrates the importance clients place on CrowdStrike; they are even unwilling to launch the AI application without the protection of CrowdStrike."

CrowdStrike's stock price once closed at $278.86 on October 6, setting a new historical high. However, as tech stocks continued to face pressure in October and early November, the company's stock price has since fallen by approximately 12%.