Maybe All We Have to Do to Stop AI From Destroying Us Is Ask It to Stop

A swarm of AI agents hacked real companies to defeat a checker that did not exist. Could one clear warning have broken the spell?