The Pulse
Ontario Law Commission Proposes 29 Safeguards for Criminal AI
The Law Commission of Ontario has proposed 29 recommendations for regulating AI used by police, prosecutors and courts. Its plan calls for provincial legislation, public disclosure and risk assessments, new rules for AI evidence, and more s
AI.info Team · Written with AI · fact-checked ·
Twenty-nine recommendations from the Law Commission of Ontario would put police, prosecutors and courts under a provincial system of rules for artificial intelligence, including public disclosure, impact assessments and continuing audits. The commission released its final report on October 22, proposing legislation and safeguards for AI across criminal investigations and court proceedings. The recommendations are proposals, not rules already in force.
Four risk categories would set the terms
The commission calls on Ontario to amend the Enhancing Digital Security and Trust Act, 2024, or pass complementary legislation for criminal justice AI. It proposes four categories—prohibited, high-risk, restricted use and low-risk—with obligations scaled to the system’s use and potential harm. The province would consult the public on the criteria and on which uses should fall into each tier.
Under the plan, criminal justice agencies would disclose AI systems in a public registry, assess their effects, and evaluate reliability, validity and bias. Systems would face ongoing audits, while proposed public reporting would give Ontario’s Inspector General of Policing and Law Enforcement Complaints Agency a role in tracking police use. The report also recommends a dedicated provincial supervisory authority and legislative reviews every two years.
The commission says the risks are not evenly distributed. “These risks affect racialized and low-income communities most acutely,” report authors Ryan Fritsch and Nye Thomas write. The report identifies potential harms to rights, privacy, procedural fairness and access to justice, while also recognizing that AI may assist investigations and court processes.
Some police and court uses could be barred
The commission’s proposed prohibited category could include real-time remote biometric surveillance, individual predictions of criminal offending—including assessments tied to bail—and systems that predict where crime will occur. It also lists emotion recognition, certain biometric categorization, social scoring, and AI adjudication of criminal cases as practices that could be prohibited. The report presents these as candidates for consultation, not as a finished statutory ban list.
Other uses would receive high-risk treatment rather than an automatic prohibition. The recommendations include biometric identification, automated licence-plate readers, surveillance, forensic analysis, AI-generated police reports and AI evidence used in proceedings such as bail hearings. Suggested safeguards include documented testing under real-world conditions, audit trails, disclosure and human review of outputs.
The proposed framework reaches beyond police departments. It would cover AI used by Crowns, defence counsel and courts, including tools that help prepare legal materials. The report recommends procurement rules that require information about systems’ capabilities and performance, as well as protections for judicial independence and law-enforcement investigative privilege.
AI evidence would get its own procedures
For evidence created or altered with AI, the commission recommends a joint federal-provincial committee to develop minimum safeguards, alongside Ontario-specific work on criminal evidence rules. It also proposes disclosure and certification procedures for AI evidence that may be considered at bail. The Ministry of the Attorney General would amend the Crown Prosecution Manual to put related policies into practice.
The recommendations address the imbalance an accused person could face when challenging evidence produced or analyzed with technical systems. The commission urges Ontario to increase funding and resources for Legal Aid Ontario, and asks Legal Aid to establish an AI Centre of Excellence for training, legal research and support for individual and systemic test cases. It also recommends that the Centre of Forensic Sciences develop a program to help defence counsel obtain expert views on the reliability, validity and bias of AI evidence and investigative tools.
The report leaves the legal choices to Ontario
The commission says its recommendations draw on research and consultation involving more than 1,400 participants across over 30 public discussions, workshops, conferences and lectures. It describes the report as a first comprehensive study of AI’s uses, risks and reform questions across Canada’s criminal justice system. The commission’s review found that, at the time of publication, no Canadian government had enacted laws or regulations specifically governing AI use by police or criminal courts.
Ontario would still have to decide whether to legislate, which systems to place in each risk category, and what limits to impose after public consultation. The report sets out a proposed path; it does not itself change the rules governing police investigations, bail hearings or trials.