A similar, software based debt-recovery occurred in Australia from 2016-2020 [1]. The government of the day in 2015 figured they could recover ~$1.5B by automating some of the processes, since there were not enough people to investigate all the possible instances. The main issue was that the software scheme used averaging, which in some cases was not accurate.
As in this story, despite people immediately claiming issue with some of the methods, it was pursued for years. And again, in several cases it tragically resulted in suicides.
In the end, after ~$1.8B of repayments and a royal commission, a couple of people resigned but there was - at least in my view - no significant repercussions for the decision makers and responsible parties at the time.