CrowdStrike Launches AI Kill Switch on Falcon Guardian Platform
CrowdStrike announced the launch of an AI kill switch in the Falcon Guardian platform, which allows security personnel to terminate the operation of enterprise AI agents in real time, prevent them from restarting, and cut off their access to the corporate network when they deviate from expectations, behave maliciously, or go out of control. The feature is similar to a circuit breaker; after agent activity is detected, a human still decides whether to cut off and restore access. CrowdStrike also announced plans to acquire AI observability startup Fiddler AI and AI security testing company Dreadnode.
Why read this
AI score 58
The original text explains the trigger conditions for the kill switch and the human decision-making process, and also highlights the real-world controversy of cross-system shutdowns.
Topics
Body · AI translation
Article body EM_StockImg_Start
EM_StockImg_End
As artificial intelligence (AI) triggers global concerns, the globally renowned cybersecurity giant CrowdStrike has announced the launch of an "AI kill switch." After the news was announced, its stock price rose more than 1% on Tuesday.
On Tuesday local time, CrowdStrike said that the company has developed an "AI Kill Switch" in its Falcon Guardian platform, which can, when AI agents within an enterprise deviate from expectations, exhibit malicious behavior, or act out of control, be used by security personnel to terminate their operation in real time, prevent them from restarting, and cut off their enterprise network access permissions.
"AI security is key to fully leveraging the advantages of AI. Deploying agents with a kill switch allows you to maintain control," said George Kurtz, CEO and founder of CrowdStrike, in a press release.
Daniel Bernard, Chief Business Officer of CrowdStrike, said, "Having the ability to control and shut down, to be able to make decisions and decide (whether to shut down), I think this is crucial for the next stage of AI applications."
A recent string of high-profile cyberattack incidents has added urgency to calls for establishing some kind of emergency shutdown mechanism. The "Hugging Face intrusion incident" orchestrated by an out-of-control OpenAI agent in July especially alarmed the business community. Last week, Binance disclosed the largest cryptocurrency cyberattack on record; hackers also recently attacked hundreds of refrigerators and freezers in French supermarkets.
This summer, U.S. lawmakers proposed a bipartisan bill requiring AI companies to build shutdown capabilities into their models. But critics argue that implementing an emergency stop button across multiple systems, data centers, and complex AI models is far more complex than it is realistic.
Bernard pointed out that in systems where customers use its Falcon platform, the kill switch works like a circuit breaker. The process can detect the activity of internal agents, but a human decision is required on whether to cut off the agent's access permissions and restore them.
"If you terminate a process at the root directory, and we have root directory access... the result is very either-or," he added: "Either you terminate the AI and the agent, or you don't terminate them, and then you report the result."
In the AI era, enterprises are rapidly adopting autonomous tools but often lack the necessary security safeguards and protections. CrowdStrike believes its new tool will enable companies to promptly stop AI agents' "misbehavior" while leaving room for innovation.
In addition, CrowdStrike also announced plans to acquire AI observability startup Fiddler AI and AI security testing company Dreadnode. To meet growing security demands, CrowdStrike and its competitors are rapidly acquiring cybersecurity startups. Demand for AI security tools has also driven a broader wave of cybersecurity mergers and acquisitions.
(Source: Cailian Press) Body middle Embedded ad