So we make a law that the CEO is responsible for actions of any agent created or operated by anyone in their company. CEOs will get serious about AI security real quick. Honestly we need to do something. There needs to be a single wringable neck.