Lawsuit says AI tool disproportionately places Black prisoners in maximum security in Ontario

An artificial intelligence program used to classify prisoners' security levels in Ontario jails is disproportionately placing Black prisoners in maximum-security conditions, according to an investigation published by The Breach and a class action lawsuit filed against the province.

The reporting, by journalist Desmond Cole, examined the Security Assessment for Evaluating Risk (SAFER) tool, which has been used in Ontario jails since 2021 to assign prisoners a risk score based on factors including arrest and disciplinary history.

Citing government data analyzed by University of Toronto criminologist Scot Wortley, the outlet reported that Black people make up roughly 5 percent of Ontario's population but represented close to 27 percent of prisoners in maximum security between 2022 and 2025, while white people were significantly underrepresented in that category relative to their share of the population.

The class action, filed by law firm Koskie Minsky, alleges the province was aware the tool's reliance on historical justice-system data — which reflects documented racial disparities in policing and charging — could disproportionately harm Black prisoners, and that Ontario adopted mitigation measures for Indigenous prisoners without extending similar safeguards to Black prisoners. Internal ministry training documents cited by The Breach acknowledged that "assessments like SAFER would likely contribute to the overrepresentation" of Indigenous and racialized people in maximum security.

The Breach reported that neither Ontario's Ministry of the Solicitor General nor Grant Duwe, the researcher credited with designing SAFER, responded to requests for comment on the program's methodology or accuracy. An Ontario Ombudsman report referenced in the investigation noted it had received complaints about the tool's disproportionate impact on Black and Indigenous prisoners, and that a ministry review of the program is ongoing.