Australian Robodebt Scheme: $1.8B Class Settlement for AI-Driven Welfare Errors
Australia's Robodebt scheme — which used automated income-averaging to issue erroneous welfare debt notices — resulted in a federal class settlement of A$1.8B and was linked to at least three suicides.
Australia's 'Robodebt' Online Compliance Intervention scheme (2015-2019) used automated income-averaging to compare welfare recipients' reported income against Australian Tax Office records — frequently fabricating debts that did not exist by inappropriately averaging annual income across fortnightly reporting periods. Over 500,000 erroneous debt notices were issued; at least three suicides were linked to Robodebt distress in evidence to the Royal Commission. The Federal Court approved a A$1.8 billion class-action settlement in June 2021 (Prygodicz v Commonwealth). A Royal Commission report in July 2023 concluded the scheme was 'a costly failure of public administration, in both human and economic terms' and referred multiple senior officials for civil and criminal investigation. Robodebt is the most-cited cautionary case in Australian government AI policy.
Systems & Vendors Implicated
What EvalGuard would have caught
Every entry in this catalogue traces back to a guardrail class — chatbot self-harm detection, facial recognition validation thresholds, deepfake watermark verification, algorithmic bias auditing, or compliance gating. See our product catalogue for the specific tools that ship those safeguards today.