Security Weekly - A CRA Resource: Block Agents or Make Them Ask

AI agent policies can do more than simply allow or block an action. They can require the agent to ask the user for explicit approval before proceeding.

That creates a human decision point inside the agent’s workflow. The agent can identify what it wants to do, but the action does not proceed until its owner approves it.

The approach offers a middle ground between unrestricted autonomy and completely blocking an agent from acting.

As AI agents become capable of taking more consequential actions, where should human approval be required?

Subscribe to our podcasts: https://securityweekly.com/subscribe

#AIAgents #AIsecurity #SecurityWeekly #Cybersecurity #InformationSecurity #AI #InfoSec