India, Nov. 4 -- AI agents are being given access to systems, applications, and networks to carry out tasks with less human involvement. But when an agent moves beyond its intended behaviour, enterprises need a way to step in. CrowdStrike's AI Kill Switch, introduced in Falcon Guardian, is designed to give security teams that control.

The feature allows an authorised security decision-maker to terminate a running agent, prevent it from launching again, and cut off its enterprise network access. It also records every action in an audit trail.

AI safety becomes a condition for wider adoption AI agents can work across endpoints, cloud environments and SaaS applications, using system-level access to complete multi-step tasks. However, they ...