Reimagining failed automation: From neoliberal punitive automated welfare towards a politics of care

This chapter argues that government turns to automated welfare benefits fraud detection must be viewed as shaped by their neoliberal and austerity driven social policy contexts. We focus on the policy lessons that can be learned through an analysis of two harmful automated fraud detection applications: the Online Compliance Intervention in Australia and the Michigan Integrated Data Automated System in the United States. Automation of fraud detection in these cases led to error, wrongful presumptions of guilt, significant burdens, stress and anxiety, and harms to individuals and families. Policy lessons include the need for a design justice approach, expert and external critique, rigorous review, equalities impact assessments, greater attention to care and community as well as rights of refusal. We argue for a move from punitive automated welfare to an approach informed by a politics of care that prioritizes peoples’ needs and promotes health and well-being while recognizing our interdependence.