Throughout history, every era has developed new technologies to organize power. Rarely have these innovations been neutral. Rather, they have often served to strengthen existing systems of domination while masking inequality beneath the language of science, efficiency, and objectivity. During the transatlantic slave trade, racial classification systems transformed African humanity into commercial property. Colonial administrations relied upon censuses, surveillance, and bureaucratic record-keeping to govern conquered populations. In the nineteenth and twentieth centuries, pseudoscientific theories of race, intelligence testing, actuarial statistics, and criminology all claimed to provide objective methods for understanding human behaviour while reinforcing anti-Black ideologies that justified unequal treatment under the law.

Artificial intelligence is the newest chapter in this long history.

Although AI is routinely promoted as impartial, data-driven, and free from human prejudice, it is neither born in a social vacuum nor insulated from historical injustice. Algorithms do not create knowledge independently; they learn from data produced by human institutions. When those institutions have been shaped by generations of systemic anti-Black racism, the algorithm inevitably absorbs those inequalities and reproduces them in new forms. Far from eliminating discrimination, artificial intelligence can automate it, granting historical injustice the appearance of mathematical certainty.

Ontario's correctional system illustrates this disturbing transformation.

Since 2021, provincial correctional institutions have quietly relied upon an artificial intelligence tool known as the Security Assessment for Evaluating Risk (SAFER) to determine the security classifications of incarcerated people. The program analyzes an individual's criminal justice history, institutional disciplinary records, prior charges, and other correctional information to generate a numerical risk score that determines whether someone is placed in minimum-, medium-, or maximum-security custody. These classifications are not simply administrative categories. They determine where a person lives, how often they may interact with others, their access to education and rehabilitation, opportunities for employment, family visitation, freedom of movement, and, ultimately, their prospects for successful reintegration into society.

A proposed class action lawsuit filed in 2025 alleges that the SAFER algorithm systematically assigns Black prisoners to harsher security classifications than other incarcerated people, resulting in significantly more restrictive living conditions and diminished access to rehabilitative opportunities. According to the claim, the Province of Ontario implemented and continued using the system despite evidence that its predictive model relied upon historical criminal justice data already distorted by decades of systemic racial discrimination. Rather than correcting institutional bias, the lawsuit argues, the technology transforms that bias into algorithmic prediction, allowing anti-Black racism to be reproduced under the authority of artificial intelligence.

The implications extend far beyond Ontario's prisons. For Black scholars, these developments cannot be understood merely as questions of technology or administrative reform. They belong to a much older genealogy of anti-Black governance. Across the African diaspora, institutions have repeatedly claimed neutrality while constructing systems that disproportionately regulate Black bodies. Slave patrols evolved into modern policing. Pass laws evolved into surveillance systems. Racial science evolved into actuarial risk assessment. Today, predictive algorithms inherit this legacy by transforming historical patterns of discrimination into statistical forecasts of future behaviour. What once depended upon openly racist ideologies is now increasingly justified through computational models, proprietary software, and the language of machine learning.

This continuity reveals one of the defining paradoxes of artificial intelligence. The more sophisticated predictive technologies become, the easier it is to mistake algorithmic output for objective truth. Yet algorithms possess no independent moral judgment. They simply identify patterns within historical data. If that history reflects disproportionate policing, discriminatory charging practices, unequal sentencing, racial profiling, and differential institutional discipline, then the algorithm learns not criminality itself but the racialized decisions embedded within the criminal justice system. In this sense, artificial intelligence does not predict future risk as much as it predicts the continuation of historical inequality.

The consequences for Black prisoners are profound. Higher security classifications frequently result in greater isolation, stricter confinement, fewer educational opportunities, reduced access to culturally relevant programming, diminished family contact, heightened surveillance, and significantly fewer opportunities to prepare for life after release. These are not merely inconveniences of incarceration; they shape mental health, rehabilitation, community reintegration, and long-term life outcomes. When such consequences disproportionately burden Black prisoners because of algorithmic decision-making, artificial intelligence ceases to be a neutral administrative tool and instead becomes an instrument through which systemic racism is reproduced at digital speed.

History teaches that systems of oppression rarely disappear. They evolve. Each generation develops new institutions, new bureaucracies, and new technologies that promise fairness while preserving familiar hierarchies of power. Artificial intelligence should therefore be understood not as a break from the history of anti-Black racism but as one of its newest expressions. Like the racial sciences that preceded it, AI derives authority from claims of objectivity while remaining deeply dependent upon the historical conditions from which its data are drawn.

It is clear that the controversy surrounding Ontario's SAFER program is not simply about flawed software. It is about the digital transformation of structural racism. Situating predictive artificial intelligence within the longer history of anti-Black surveillance, criminalization, and state control, it contends that algorithmic governance represents one of the most significant civil rights challenges of the twenty-first century. As governments increasingly entrust machines with decisions that profoundly shape human lives, the central question is no longer whether artificial intelligence is efficient. The question is whether democratic societies will permit centuries of racial inequality to be encoded into algorithms and administered as objective justice—or whether they will insist that technological innovation be guided by historical truth, transparency, accountability, and racial justice.