There’s a big disconnect between the rate at which IT is deploying various kinds of AI-generated output and the speed with which it can roll those changes back when things go wrong, warns a new report from Fleet Device Management.