I honestly don’t know what anyone expected given all the labs started building self-driving malware at about the same time.

I remember lots of people talking about running agents in VMs or on whole dedicated machines, yet barely anyone talking about how they were using mundane OS access controls.