Comment by mazieres

What would it take for people to stop recklessly running unconstrained AI agents on machines they actually care about? A Stanford researcher thinks the answer is a new lightweight Linux container system that you don't have to configure or think about.