Algorithm Assessing Risk Faced by Victims of Gender Violence Misclassified Low-Risk Cases, Allegedly Leading to Homicide of Women and Children in Spain

VioGén is an algorithmic risk assessment system deployed by Spain's Ministry of Interior in 2007 to evaluate the likelihood of repeat domestic violence incidents. The system uses a questionnaire with 39 items (in version 4.0) that police officers complete with victims to generate risk scores ranging from 'no risk appreciated' to 'extreme risk.' The algorithm determines what level of police protection women receive after reporting abuse. Multiple audit reports and investigations found severe flaws in the system. In 2014, 14 out of 15 women who were killed after reporting their abusers had been classified as 'low risk' or 'no risk appreciated' by VioGén. The system has been criticized for lack of transparency, minimal human oversight (officers only modify the AI's risk assessment in 5% of cases), systematic underestimation of psychological violence, and adapting risk classifications based on available police resources rather than actual danger levels. An external audit by Eticas Consulting found that in 2021, only 1 out of 7 women who sought police protection actually received it, with approximately 45% of cases receiving 'no risk appreciated' classifications. The system processes over 600,000 cases and has been updated multiple times, most recently in 2019 with version 5.0 that includes specific assessments for lethal violence risk.