FIJISHI EPL

Canada's Treasury Board Directive on Automated Decision-Making first issued April 1 2019 required existing automated decision systems to comply with updated requirements by June 24 2026. The Federal AI Register operational since November 2025 contains 409 automated decision systems across 42 departments with ontology emphasising system capabilities and procurement lineage but omitting structured fields for human-in-the-loop configuration. The Directive is not law and applies to approximately 100 departments under the Financial Administration Act. The Algorithmic Impact Assessment tool is an online questionnaire producing a risk score without cryptographic proof of correct autonomous operation. The fourth Directive review is currently underway. No version of the Directive specifies the constitutional command architecture that makes compliance with transparency accountability and fairness obligations cryptographically verifiable in a federal court proceeding. This paper documents the structural gap between Canada's world-leading federal AI governance compliance culture and the constitutional command standard required to make that compliance legally defensible.

Read paper: https://doi.org/10.5281/zenodo.20739113
For full paper and sovereign briefings: protocol@fijishi.com