High-impact system is the term Bill C-27’s proposed Artificial Intelligence and Data Act (AIDA) used for the category of AI systems that would have triggered risk-mitigation duties — and, notably, the bill did not define which systems counted in its own text at all.
As introduced at first reading, the bill’s definitions section stated only that “high-impact system means an artificial intelligence system that meets the criteria for a high-impact system that are established in regulations” — leaving the actual criteria to be written later, in regulations that were never made.
As recorded on LEGISinfo, read 27 August 2026, Bill C-27 is shown against the 44th Parliament, 1st session (22 November 2021 to 6 January 2025), a session the page itself marks as “a prior session”, with its status listed as “At consideration in committee in the House of Commons” and its latest recorded activity second reading and referral to committee on 24 April 2023. Nothing about the bill’s force should be asserted beyond that sentence.
“High-impact” is not the same idea as “high-risk.” The European Union’s AI Act uses “high-risk” for its own, separate risk-tier system: “Most of the text addresses high-risk AI systems, which are regulated”, as artificialintelligenceact.eu summarizes it (an unofficial consolidation of the EU text, cited because it is readable; the official text is published on EUR-Lex). The two regimes are not interchangeable: AIDA never became law and never fixed which systems its term covered, while the EU regime is a different jurisdiction’s own classification with its own criteria.
A Canadian business trying to decide today whether its AI-driven credit-scoring tool would have been a “high-impact system” under AIDA cannot actually answer that question — not because the answer is unclear, but because the bill itself never wrote it down. The criteria were left to future regulations, and no such regulations were ever finalized before the bill stalled in committee.
That is a materially different situation from the EU AI Act, where a business can at least read a settled classification system to find its own answer, or from Canada’s own Treasury Board directive, where “Impact Assessment Levels I–IV” are defined and in force today — just only for federal government systems, not this business’s own tool.
See also: algorithmic impact assessment, what AIDA proposed and where it stands, AI governance.
Judging how much scrutiny a given AI system actually warrants, in the absence of a settled Canadian classification, is exactly the kind of assessment ai-due-diligence works through case by case.
