Case-files

In-depth forensic records of individual systems. Each is researched by hand from public records — at least three primary sources (courts, regulators, official inquiries) and three secondary sources, with every key claim confirmed by two independent sources — and the harm is coded on seven dimensions, which are never added together. See Method.

• S-001 · Fixed rulesSyRIDutch state inter-agency risk-scoring of welfare recipients, halted by the District Court of The HagueNetherlands · 2014–2020 · Concluded (court halt 2020); records persisted after the halt
• S-004 · Fixed rulesToeslagenaffaireDutch childcare-benefits fraud classification that wrongly accused tens of thousands of familiesNetherlands · 2004–2020; remedy ongoing · Remedy ongoing and incomplete (compensation operation targeted for completion in 2027)
• S-005 · AI (machine learning)Optum Impact ProCommercial healthcare risk algorithm that under-estimated the health needs of Black patientsUnited States · 2013–2015 (audited sample); product still marketed in 2026 · Unresolved — correction verified on a replication dataset only; no enforcement
• S-007 · Statistical modelCAF/CNAF scoring algorithmFrench family-benefits risk score that prioritises welfare recipients for fraud controlsFrance · 2010–present (pilot 2004) · Active — challenge pending before the Conseil d'État; system in operation
• S-008 · Statistical modelSCHUFA credit scoreGerman private credit-scoring system held by the Court of Justice of the EU to be an automated decisionGermany (EU-wide reach via the CJEU) · Coded 2018–early 2026 (pre-reform baseline) · Partially resolved — CJEU 2023 rulings; 2026 reform announced, effect unverified
• S-009 · Fixed rulesRobodebtAustralian automated debt-raising scheme found unlawful by the Federal Court and a Royal CommissionAustralia (Commonwealth) · July 2016 – November 2019 · Concluded — unlawful; settlements paid; deaths recorded on the public record are irreversible