Algorithmic Bias: How Ontario’s AI Prison Tool Targets Black Inmates
A growing legal and ethical controversy has emerged surrounding the use of artificial intelligence within Ontario’s correctional system. Recent findings suggest that the software tasked with determining inmate security classifications is systematically funneling Black prisoners into maximum-security environments at rates far exceeding their demographic representation.
### The SAFER Tool Under Scrutiny
At the heart of the issue is the Security Assessment for Evaluating Risk (SAFER) system. Implemented by the province in 2021, this digital tool was designed to standardize the classification process by analyzing variables such as prior disciplinary records and arrest history. However, critics argue that by relying on historical data-which often reflects long-standing systemic biases in policing-the AI effectively automates and amplifies racial discrimination.
### Statistical Disparities in Incarceration
The scale of this disparity is stark. According to an analysis conducted by University of Toronto criminologist Scot Wortley, the data reveals a profound imbalance. While Black individuals account for approximately 5 percent of the total population in Ontario, they comprised nearly 27 percent of the maximum-security inmate population between 2022 and 2025.
This data, which serves as a cornerstone for a pending class-action lawsuit against the province, highlights the dangerous “feedback loop” created by predictive policing and sentencing tools. When an algorithm is trained on data sets that already contain historical prejudices, it inevitably produces skewed outcomes that disproportionately penalize marginalized communities.
### The Human Cost of Automated Justice
The investigation, spearheaded by journalist Desmond Cole, underscores a critical flaw in the “neutrality” of technology. By delegating life-altering decisions-such as where an inmate is housed and what level of freedom they are afforded-to an opaque software program, the justice system risks shielding itself from accountability.
As the class-action lawsuit moves forward, it raises fundamental questions about the role of AI in public institutions. If a tool intended to provide objective risk assessment instead functions as a mechanism for racial profiling, it challenges the very integrity of the correctional process. The case serves as a cautionary tale for other jurisdictions globally that are increasingly turning to automated systems to manage public safety and judicial outcomes.
