Strong agree! Auditing, transparency, and accountability in algorithmic systems will (hopefully) see much more study and policy in the next decade...
It's rare that academic terms so perfectly capture the issue:
Humans-in-the-loop of an automated system risk being the 'moral crumple zone', like a car bonnet 'designed to absorb the force of impact in a crash', suffering the 'moral and legal penalties when the system fails'