AI Risk‑Assessment Tool Accused of Racial Bias in Ontario Prisons

When you think about artificial intelligence, you probably picture futuristic robots or clever chatbots. But the same technology is also deciding how dangerous a person is deemed to be behind bars – and, according to a recent investigation, it might be doing a terrible job when it comes to race.

The program in question is called SAFER – the Security Assessment for Evaluating Risk – and it’s been the backbone of Ontario’s prison‑security decisions since 2021. In theory, SAFER crunches data about an inmate’s arrest record, disciplinary history and a few other factors, then spits out a risk score that determines whether someone lands in a regular wing or the maximum‑security block.

Sounds straightforward, right? Not quite. A deep‑dive by journalist Desmond Cole for The Breach, backed by data from the province and analysis by University of Toronto criminologist Scot Wortley, shows a glaring disparity. Black Ontarians make up about 5 percent of the province’s total population, yet they accounted for roughly 27 percent of the prisoners placed in maximum‑security facilities between 2022 and 2025. White inmates, by contrast, were noticeably under‑represented in those high‑security cells.

That gap isn’t a random statistical blip. The class‑action suit filed by Koskie Minsky argues that Ontario officials knew the SAFER tool leaned heavily on historical criminal‑justice data – data that already reflects systemic over‑policing and over‑charging of Black communities. The lawsuit says the province rolled out mitigation measures for Indigenous prisoners, but never extended the same safeguards to Black inmates, effectively leaving a blind spot.

Internal ministry documents, which The Breach obtained, even admitted that “assessments like SAFER would likely contribute to the overrepresentation of Indigenous and racialized people in maximum security.” Yet, when reporters reached out for comment, neither the Ministry of the Solicitor General nor Grant Duwe – the researcher credited with designing SAFER – responded.

The Ontario Ombudsman has already taken note. An earlier report referenced in the investigation mentions complaints about the tool’s disproportionate impact on Black and Indigenous prisoners, and it says a ministry review is still underway. Still, families of those locked up in the toughest units are waiting for answers, and the lawsuit is pushing the province to either overhaul the algorithm or abandon it altogether.

For many, this case feels like a modern‑day echo of older injustices – a reminder that tech, without careful oversight, can amplify the very biases we hoped to leave behind. As the legal battle unfolds, the hope is that Ontario will either fix the flaws in SAFER or find a better way to keep prisons safe without sidelining entire communities.