Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor

Automating Inequality is not about a future superintelligence escaping control. It examines systems already deciding access to food, healthcare, housing, and family integrity through routine administration. Virginia Eubanks enters Indiana’s automated welfare eligibility program, Los Angeles’s coordinated system for homeless housing, and Allegheny County’s child-welfare risk model. Through the experiences of people subjected to them, she shows that algorithms do not remove politics from decisions. They encode scarcity, suspicion of poverty, and punitive help into a new technical process.

Indiana demonstrates how “efficiency” can become a machine for denying service. After the state transferred benefits administration to private contractors and remote systems, a missing form, missed call, or failed document transfer could be classified as “failure to cooperate.” About one million applications for healthcare, food, and cash assistance were denied in three years. Errors were no longer repaired by a caseworker familiar with a household but dispersed across call centers, databases, and appeals. Automation reduced not poverty but the institution’s obligation to meet applicants and explain itself.

Los Angeles confronts a reality in which housing is far scarcer than need. A questionnaire converts health, trauma, violence, and street experience into a vulnerability score used to rank people for limited placements. Ranking may be more consistent than improvisation, but it cannot create a home. It also requires people to exchange intimate histories for a position in the queue. Someone with a high score may still wait years, while a low score can mean exclusion for not being vulnerable enough. Technology translates the political question—why is there not enough housing?—into finer management of shortage.

Child protection reveals how data can confuse poverty with danger. A risk model draws on families’ histories of public-service use. Low-income households necessarily encounter welfare, health, and justice agencies more often and leave more computable records. Affluent families buy private care and remain comparatively data-poor; poor mothers, Black families, disabled parents, and single-parent households become more visible, reportable, and investigable. Prediction can then form a feedback loop: surveillance creates records, and records become evidence that more surveillance is needed.

Eubanks calls this structure a “digital poorhouse,” connecting it to the longer U.S. history of poverty governance. County poorhouses, eligibility tests, and scientific charity distinguished the deserving from the undeserving in order to restrict aid. Data matching, fraud detection, and risk prediction inherit that ethic while making exclusion appear objective. The feminist stakes are direct because care needs are gendered. Mothers, caregivers, disabled people, and survivors of violence rely more heavily on public systems and may be scrutinized precisely for seeking help. Antifraud policy recodes survival strategies as moral failure.

The book does not claim that technology is inherently evil or that human bureaucrats are naturally fair. Transparent, appealable systems supported by adequate resources can reduce some arbitrariness. The problem is automation placed inside austerity and punishment, then tested first on people with the least power. Three U.S. cases cannot represent every country or later model, and systems and laws change. The durable question is not a particular parameter but whether technology expands resources and agency or makes refusal, surveillance, and family separation cheaper.

For FemRes readers, the book offers a concrete public-technology test. Does a system address need or merely rank it? Do people know which data are used and can they understand and challenge a decision? Who may opt out, and who must surrender privacy to survive? Who repairs errors, and onto whom are savings transferred? Read with welfare-rights organizing, family-policing abolition, disability justice, and care-economy scholarship. Without distributive justice, a smarter allocation system only manages inequality more smoothly.