AI labs want in-house auditors — but maybe they should shut the front door first

There may be a simpler and more effective fix for rogue agents, hiding in plain sight.

from TechCrunch https://ift.tt/aMqzF0G