Registry
Short, sourced records of AI and other automated systems that caused documented harm. Each record was found by the collection pipeline and then checked by the author against its sources; only checked records are published (76 more are in review). A registry record is a documented incident, not a full forensic case-file. Records are listed without ranking or scores.
17 of 17 records
- R-0099RisCanvi risk scores shape prisoners' access to leave and paroleCriminal justice · Rule-based risk scoring
- R-0132Essex Police pauses live facial recognition over biasPolicing · Face recognition (AI)
- R-0138Facewatch flags a shopper as a shoplifter in errorOther · Face recognition (AI)
- R-0149Met Police live facial recognition: mass scanning and misidentificationPolicing · Face recognition (AI)
- R-0458Gaggle flags a student's unsent draft emailEducation · Machine learning (AI) classifier
- R-0524Licence-plate network searched to track a woman after an abortionPolicing · Other automated system
- R-0570Meta's automated moderation removes lawful abortion informationContent moderation · Machine learning (AI) classifier
- R-0769AI-assisted social-media screening used to revoke student visasImmigration · Other automated system
- R-0943Face recognition searches lead to wrongful arrestsPolicing · Face recognition (AI)
- R-1029OASys risk scores inform bail, sentencing and paroleCriminal justice · Rule-based risk scoring
- R-1976Face recognition match leads to Robert Williams's wrongful arrestPolicing · Face recognition (AI)
- R-2168Rite Aid face recognition falsely flags customersOther · Face recognition (AI)
- R-2579Google's image scanning reports parents for medical photosContent moderation · Machine learning (AI) classifier
- R-3573ShotSpotter alerts used to justify stops and frisksPolicing · Machine learning (AI) classifier
- R-3703Rikunabi sells predictions of which students will decline job offersHiring · Machine learning (AI) classifier
- R-3764Pasco Sheriff's scoring targets young people for repeated home visitsPolicing · Rule-based risk scoring
- R-4572YouTube's automated flagging removes Syrian human-rights evidenceContent moderation · Machine learning (AI) classifier