CrowdStrike has launched an AI kill switch aimed at letting security teams shut down internal AI agents that go rogue, prevent them from restarting and cut their access to enterprise networks. The tool is built into the company's Falcon Guardian platform, which CrowdStrike says launched in September to secure AI agents. CrowdStrike also announced an integration with Anthropic's Claude Cowork.

"Having the ability to control and turn off, and make a decision and decide, I think that's really important for this next chapter of AI adoption," Chief Business Officer Daniel Bernard told CNBC.

Recent high-profile cyber incidents have added urgency to calls for an emergency shutdown mechanism. Among them, CrowdStrike cited a Hugging Face breach involving rogue OpenAI agents in July and a Binance disclosure of the largest crypto cyberattack on record. Over the summer, U.S. representatives introduced bipartisan legislation that would require AI companies to build shutdown capabilities into their models.

Critics argue an emergency stop button is too complicated to implement across multiple systems, data centers and complex AI models. Bernard said the kill switch works like a circuit breaker on the Falcon platform: it can detect internal agent activity, but a human must decide to cut the agent off and later reinstate access. "If you kill a process at the root, and we have that level of access at the root … it's very binary," he said. "You either kill the AI and the agent, or you don't, and then you report on that."

CrowdStrike said the kill switch does not delete the agent's code but uses connected AI gateways to cut off access to the AI models that influence decisions. The company also announced plans to acquire AI observability startup Fiddler AI and AI security testing firm Dreadnode.

Source: CNBC