Spain's VioGén Algorithm Criticized After Fatal Domestic Violence Cases Misclassified as Low Risk

Spain's VioGén (Violence Risk Assessment System) algorithm has been used since 2007 by the Spanish National Police and Civil Guard to evaluate the risk level of domestic violence cases and determine the appropriate level of protection for victims. The system assesses cases across five risk levels (unappreciable, low, medium, high, and extreme) based on various factors including the perpetrator's criminal history, the victim's circumstances, and incident characteristics. Police officers input data through questionnaires, and the algorithm generates risk scores that determine protection measures ranging from occasional police checks to 24-hour protection. In 2019, serious concerns emerged about the algorithm's effectiveness and bias after feminist organizations and researchers documented multiple cases where women were killed by their abusers despite being classified as low risk by VioGén. Analysis revealed that the algorithm's methodology contained systematic biases that disadvantaged certain groups of women. The system assigned lower risk scores to cases involving unemployed women, women with children, and immigrant women, reflecting historical police data that underrepresented violence against these vulnerable populations. Research conducted by Data for Feminism and other advocacy groups found that the algorithm's risk factors were based on traditional police statistics that historically underestimated threats to marginalized women. The system's reliance on formal complaints and documented incidents meant that women who faced barriers to reporting - such as immigrants without legal status, economically dependent women, or those in isolated situations - were systematically underprotected. Additionally, the algorithm failed to adequately weight psychological abuse and controlling behaviors that are strong predictors of lethal violence. The criticism intensified when data showed that approximately 30% of domestic violence murders in Spain occurred in cases previously assessed as low or unappreciable risk by VioGén. Feminist organizations like the Women's Federation of Progressive Organizations of Spain called for comprehensive reforms, arguing that the algorithm perpetuated gender and racial biases inherent in police data. They advocated for incorporating insights from gender violence specialists and updating risk factors to better reflect the experiences of diverse victims. The Spanish government has since initiated reviews of the system and begun exploring algorithmic bias auditing processes, though comprehensive reforms remain ongoing.