Coolcaesar / Wikimedia Commons | CC BY4.0 | Archive,2024
CrowdStrike introduced an AI Kill Switch on Tuesday, giving security teams a way to stop an AI agent that strays from its intended behavior, prevent it from restarting and cut off its enterprise network access, the company said.
The feature is part of Falcon Guardian. It is available now at the gateway, while endpoint support is coming “soon,” according to CrowdStrike’s announcement. The company did not give a date for the endpoint rollout and said unreleased features remain in development and are subject to change.
An authorized administrator triggers the action, rather than leaving the final decision to the agent. Each action is recorded in an audit trail, CrowdStrike said. Its policy controls can also block the agent from relaunching on a selected host or group of hosts.
“AI safety is how you get the full benefit of AI. When you deploy agents with a kill switch, you stay in control,” CEO and founder George Kurtz said in the release.
The launch addresses a practical problem for businesses deploying agents with access to company systems: an agent can take an unauthorized action even without an outside attacker, CrowdStrike said. The new controls add a human stop mechanism to Falcon Guardian’s existing agent discovery, policy enforcement and attack detection tools.
The announcement follows CrowdStrike’s signed agreements to acquire Dreadnode and Fiddler AI, which have not closed. The Wall Street Journal put the combined purchase price at $495 million, according to Yahoo Finance.
JBizNews Desk | Austin, Texas
© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.
