VioGén was launched in 2007 by Spain's Interior Ministry under Organic Law 1/2004 on Comprehensive Protection Measures against Gender Violence, and updated to VioGén 5.0 in 2019 with an AI pilot to refine the underlying model beginning in 2022. It runs nationwide except in the Basque Country and Catalonia, which operate their own separate systems.

Officers complete two structured questionnaires: the Police Risk Assessment (VPR), used at first report, and the Police Risk Evolution Assessment (VPER), used to reassess a case over time. Together they cover 35 risk indicators. The output is a weighted score sorting the case into one of five bands -- unappreciated/negligible, low, medium, high, or extreme -- and that band determines the level of police protection assigned to the victim. The Eticas Foundation, which conducted an independent audit, describes the underlying method as classical statistical weighting rather than machine learning, a distinction worth being precise about on a site that otherwise covers AI specifically.

No dedicated external regulator has been identified. Eticas offered a confidential pro-bono audit from 2018 and, after no action followed, conducted its own 2021 audit by reverse-engineering the system with the Ana Bella Foundation, finding that officers keep the system's automatic outcome in 95% of cases despite the score being designed as a recommendation an officer may override. Reporting by the New York Times found that at least 247 women were killed by partners after being assessed by VioGén since 2007, of whom at least 55 had been scored negligible or low risk. As of 2025 the system carried roughly 92,000 active cases, 83% of them classified low or negligible risk.

This is decision-support and risk assessment for protective policing, not the predictive-policing sense of forecasting where crime will occur, and is catalogued as such deliberately.