News · eGov
After consulting more than 1,400 people across Ontario, the Law Commission of Ontario has released a final report with 29 recommendations for governing artificial intelligence in criminal law.
✦ AI SUMMARY 30-second read
- The LCO report calls for a provincial criminal AI framework, new legislation, a dedicated oversight authority and a public AI registry.
- The federal government's current AI strategy does not mention criminal law or related legislation, leaving a significant policy gap.
- The report recommends banning any AI tools that make predictions about whether an individual will commit a crime.
Why it matters: Without a legal framework, Canadian police services remain free to adopt AI tools that could infringe Charter rights and disproportionately affect marginalised communities.
This summary was generated by AI. Editors may review and update it.
The Law Commission of Ontario has published its final report on artificial intelligence and criminal law, following consultations with more than 1,400 people across the province. The report contains 29 recommendations and calls on government to introduce comprehensive rules covering how AI is used across the criminal justice sector.
The project began in 2025 with a series of issue papers and consultations. Its central argument is that Canada needs its own legislative approach, one grounded in Charter rights, rather than adopting models already in use in the United States or the United Kingdom.
What was recommended
The LCO is calling for a provincial criminal AI framework supported by new legislation specific to AI and criminal law. It also recommends establishing a dedicated provincial authority to oversee AI systems and creating a public AI registry so citizens can see which tools are in use.
The report recommends an outright ban on any tools that predict whether a named individual will commit a crime in future. It also urges new evidentiary rules and procurement standards, arguing that police services currently feel powerless to negotiate standard terms of use with technology vendors.
Ryan Fritsch, LCO legal counsel and project lead, said that just as breathalyser tests required new rules around use, efficiency, evidence and procedural fairness, the same approach can be applied to AI.
The current landscape in Canada
Police forces in Toronto, Calgary, Vancouver, Ottawa and Winnipeg are already using facial recognition software, video analytics and AI-assisted police report drafting. Crown prosecutors are permitted to use pre-approved AI prompts in Microsoft Copilot for drafting legal memos and research, and courts have used AI for judgment writing, legal research and generating transcripts.
Several Canadian police services have experimented with predictive policing tools. The Calgary Police Service began using Palantir Gotham in 2016 before cancelling the contract. The Vancouver Police Department developed its own predictive policing software, GeoDASH, while the Saskatoon Police Service and the University of Saskatchewan ran a pilot tool in 2019 aimed at identifying potential missing persons.
The federal government's recent AI strategy contains no mention of criminal law or new legislation in this area. The Canadian Association of Chiefs of Police released principles for responsible use this year, but binding rules remain absent.
Lessons from the US and UK
The LCO report examined AI adoption in criminal justice systems in the United States and the United Kingdom, both of which are further ahead than Canada. In the US, the most widely used law enforcement AI tool is an automated licence plate reader system; about 140,000 such readers are currently in operation, drawing significant public pushback. More than 50 cities cancelled related contracts this year amid privacy concerns. The US government signalled in a 2025 executive order that it would pursue a minimally burdensome national AI policy, though 38 states went on to adopt their own legislation.
In England and Wales, more than 4.6 million faces were scanned by live facial recognition technology in 2024. A 2025 government report there recommended data audits, greater transparency and more human oversight. Marion Oswald, a law professor at Northumbria University involved in the PROBabLE Futures project, cautioned Canada against following the UK model of leaving procurement decisions to individual police services. She argued that clear regulation improves, rather than hinders, innovation.
Oswald also warned against AI credibility tools designed to detect deception through voice or facial expressions, noting they lack scientific validity and independent evaluation. A report by researchers from the University of Glasgow and Northumbria highlighted cascade errors in connected AI systems, including one case where a witness statement was materially altered by an AI tool in a way that changed its legal significance.
What happens next
LCO staff are continuing to consult with police services, courts, lawyers and the public. Whether the report's 29 recommendations will be translated into legislation remains an open question.
Fritsch emphasised that technology does not set the rules — government does — and expressed optimism that the recommendations could be implemented to bring meaningful change to the justice system.