01
Factual background & dispute
- The Correctional Service used five psychological and actuarial risk assessment tools to evaluate incarcerated individuals.
- The tools were developed and validated predominantly on non-Indigenous populations.
- The claimant argued that assessment results could affect security classification, parole and correctional programming.
02
Core issues & judicial focus
- Whether high-impact assessment tools must be validated against the specific demographic populations to which they are applied
- How public bodies must ensure that administrative decision-making relies on accurate, up-to-date, and complete information
- What procedural and substantive consequences arise when an assessment tool lacks validated demographic applicability
03
Judicial finding & holding
- The Supreme Court partly allowed the appeal and declared a breach of section 24(1) of the Corrections and Conditional Release Act.
- The correctional authority had long known of potential cultural bias but failed to take reasonable steps to verify the tools for Indigenous offenders.
- The Court did not allow the claims under sections 7 and 15 of the Canadian Charter of Rights and Freedoms.
- The psychometric and actuarial tools are not identified here as generative AI or machine-learning systems; the review concerns the 2018 judgment.
04
Practical risk implications
01Model validation must explicitly cover operational target populations and significant demographic sub-groups.
02Deploying organizations remain responsible for local validation of externally procured assessment tools.
03High-impact systems require ongoing updates to validation conclusions, known limitations, and remedial adjustments.