Riseline AI Draft Talk to us

AI explained

In policing, AI is software that finds patterns in data and produces an output: a transcript, a draft, a match or a map. Here is what each type does, and what to watch for.

Tools assist, people answer

Across Canadian sources, officers stay accountable for decisions made with AI help.

Each tool fails differently

Drafting, matching, transcribing and predicting each raise their own questions.

Mostly pilots so far

Reported Canadian uses are largely pilots and early deployments.

Report drafting (generative AI)

Generative AI writes text. The best-known Canadian example is Axon's Draft One, which turns body-camera audio into a draft report that an officer reviews. A report underpins bail, prosecution and civil cases, so accuracy matters most here.

Read more

The RCMP is piloting Draft One in Alberta and British Columbia. As reported by CBC, the program is powered by the same model used by ChatGPT, the RCMP says it turns down the dial on creativity to reduce factual errors known as hallucinations, and officers must change at least 10 per cent of what is produced. [CBC / Canadian Press, June 2026]

A Police1 article for supervisors describes the risk as "polished inaccuracy": language that sounds clean and professional but does not reflect the encounter. [Police1] See the Q&A on court and the supervisor checklist.

Sources

Facial recognition

Facial recognition compares an image of a face with a database and returns possible matches. Canada's privacy authorities say it can pose serious risks to privacy, and that a person should review matches before any decision to detain, investigate or charge.

Read more

The authorities also say the technology has a tendency to misidentify some gender and racial groups. [Joint guidance, May 2022] [OPC, 2021]

York and Peel Regional Police share a system for comparing crime-scene images with mug shots; a York spokesperson said a match alone would not be grounds for an arrest. [Blue Line / Canadian Press, April 2025] A contested example is Clearview AI; see Law & Rules.

Sources

Transcription and translation

Speech-to-text tools turn audio into transcripts, and translation tools move speech between languages. Accuracy can suffer with accents, slang and people talking over each other.

Read more

Edmonton Police Service has been building an in-house transcription tool, and Calgary Police Service has announced it is testing a real-time translation service. [Blue Line / Canadian Press, April 2025] [CBC / Canadian Press, June 2026] A University of Alberta law professor quoted by CBC notes nuance can be lost. [CBC / Canadian Press, June 2026]

Sources

Analytics and prediction

Analytics tools search records for patterns such as linked cases and repeat locations. Predictive tools, which forecast where or when crime may occur, draw the strongest criticism from civil liberties groups and researchers.

Read more

Lethbridge Police Service describes an in-house AI-enhanced version of CompStat and an intelligence portal; its reported results are its own claims. [Blue Line, 2025] The Canadian Civil Liberties Association has objected to some predictive techniques, and Citizen Lab's 2020 report argues algorithmic policing can engage Charter rights and embed bias. [Blue Line / Canadian Press, April 2025] [Citizen Lab, 2020] These are advocacy and research positions, not court rulings.

Sources
The human in the loop. The Toronto Police Service Board policy says members are accountable for all decisions they make, "including those made with the assistance of AI technology". [TPSB policy]

Next: Law & Rules or the checklists.

Descriptions come from agencies, journalists or vendors as reported; we have not tested any product. Last checked 2 November 2026.