AIUpdated November 3

CrowdStrike unveils an AI-agent kill switch.

Where CrowdStrike says the switch sits is the point: outside the AI model, working while an agent runs. CEO George Kurtz says it can stop any agent, and that the agent cannot argue with, rewrite or wait out the control. Decrypt describes the tool as a gateway that lets security teams cut off unauthorized agents. So this isn't a request for the model to behave; it's a control the company says stays beyond the agent's reach.

2 publishers2 reports

Coverage started November 3