Ontario Launches First Major Review of AI in Criminal Justice System

The Law Commission of Ontario (LCO) has launched a public consultation to examine the implications of artificial intelligence (AI) in the criminal justice system, marking the first comprehensive initiative of its kind in Canada. The project is intended to assess how AI is being used across all phases of criminal justice, identify risks and gaps, and guide the development of responsive regulatory frameworks.

The LCO’s “lifecycle” approach will examine the use of AI in police investigations, prosecutorial decision-making, bail and sentencing assessments, courtroom procedures, and post-conviction oversight. Particular attention will be paid to potentially controversial applications such as facial recognition, predictive policing, digital device de-encryption, and high-risk inmate identification systems.

While AI tools are increasingly being deployed in justice systems around the world—including for probabilistic DNA analysis, sentencing, and bail assessments—Canada currently has no legal framework specifically governing the use of AI by law enforcement, prosecutors, courts, or defense counsel. This regulatory gap is a key driver behind the LCO’s consultation, which aims to ensure that emerging technologies do not undermine rights to due process, access to justice, or civil liberties.

In February, the Ontario Privacy Commissioner issued guidance on the use of facial recognition by police, but recent legislation governing public sector AI in Ontario excluded courts and police services—leaving core areas of the justice system without binding oversight.

The consultation process is also responding to ongoing concerns about the potential for bias, systemic discrimination, and the disproportionate impact of AI on marginalized communities. The LCO is working with the Ontario Human Rights Commission to develop an AI impact assessment tool to help organizations evaluate whether algorithmic systems meet standards of fairness, transparency, and legal compliance.

Transparency and explainability are central themes of the review. The growing use of proprietary or “black box” AI systems has raised questions about whether defendants and their legal counsel can fully understand or challenge algorithmic outputs that may influence charges, bail conditions, or sentencing recommendations. Public reporting, algorithmic audits, and access to underlying data are among the best practices being discussed.

The LCO also highlights the need for training across the justice system, recommending that judges, prosecutors, defense lawyers, and administrators receive instruction on the technical foundations, limitations, and implications of AI. This is intended to prevent undue reliance on tools whose outputs may appear authoritative but are based on opaque or biased datasets.

Among the broader policy goals outlined by the LCO is the development of a “Trustworthy AI Framework” for high-risk technologies, including pre-deployment review processes, ongoing public consultation, and the involvement of multidisciplinary advisory groups. The Commission stresses that deployment of AI tools in criminal justice should not proceed without mechanisms for accountability, review, and public engagement.

The LCO has released a series of issue papers to support this work, and established an external Advisory Committee to provide guidance throughout the consultation. Input is being solicited from a wide range of stakeholders, including legal professionals, law enforcement, technologists, advocacy groups, and members of the public.

Sources: Law360, Law Commission of Ontario

—
April 23, 2025 – by Cass Kennedy