High-impact AI system
An AI system whose use poses significant risk of harm or rights impacts.
How the laws define this
Both frameworks agree that high-impact AI systems are defined by their potential to cause serious harms or severe risks to fundamental rights and interests, subjecting operators to heightened responsibilities such as mandatory risk assessments, safety or mitigation measures, and transparency or monitoring obligations. The primary divergence lies in how these systems are identified and classified. Under the Digital Charter Implementation Act (Bill C-27), the criteria and lists designating high-impact systems are contemplated to be specified by regulation under AIDA. Conversely, the Seoul AI-based Administration Enhancement Plan outlines specific classification criteria directly within its scope—including application area, potential risks, and risk severity—and explicitly includes systems trained with massive computing power. Operators navigating these instruments must prepare for future regulatory lists under AIDA while concurrently evaluating technical training parameters under the Seoul plan.
Synthesised from the 2 statutory definitions below. Each one is quoted in full, with its source.
How jurisdictions differ
What is distinctive in each definition, binding law in force first.
| Jurisdiction | Instrument | Status | What is distinctive |
|---|---|---|---|
| Seoul, South Korea | Seoul AI-based Administration Plan | Adopted | Adds classification criteria including application area and massive computing power alongside risks to fundamental rights and their severity, while imposing heightened responsibilities on operators. |
| Canada | Digital Charter Implementation Act (Bill C-27) — An Act to enact the Consumer Privacy Protection Act, the Personal Information and Data Protection Tribunal Act and the Artificial Intelligence and Data Act | Withdrawn | Specifies mandatory risk assessments, mitigation measures, documentation, and transparency obligations, with criteria and lists to be defined by regulation. |
What it triggers
- Under Canada's Digital Charter Implementation Act (Bill C-27), these systems would have required mandatory risk assessments, mitigation measures, documentation, and specific transparency obligations.source
Definitions (2)
An AI system identified by AIDA as posing a heightened likelihood of serious harms or material impacts to rights and interests; such systems are subject to mandatory risk assessments, mitigation measures, documentation, and specific transparency obligations. The bill contemplated that criteria and lists of high‑impact systems would be specified by regulation.
AI systems classified by criteria such as application area, potential risks to fundamental rights, and severity of those risks (including those trained with massive computing power). Operators of these systems are subject to heightened responsibilities, including risk assessment, safety measures, monitoring, and potential reporting or inspection obligations.
Not to be confused with
That term emphasizes significant potential to harm rights or safety, while these definitions detail specific classification criteria and operator obligations.
That term focuses on material effects on life, safety, or fundamental rights, whereas these include additional factors like computing power or regulatory lists.
That term highlights substantial effects on safety, health, or fundamental rights, but these definitions add operator duties and specific risk assessment processes.
Questions and answers
What would high-impact AI systems have faced under Bill C-27 in Canada?
They would have required mandatory risk assessments, mitigation measures, documentation, and specific transparency obligations. Criteria and lists would have been specified by regulation.
How do the definitions of high-impact AI systems differ between Seoul and Canada?
Seoul's definition includes application area and massive computing power, focusing on operator responsibilities in a non-binding plan. Canada's withdrawn bill emphasized regulatory criteria for assessments and transparency.
Related Terms
High‑impact AI system
AI systems posing significant risk to people or society....
High-impact / High-risk AI system
AI systems with significant potential to harm rights or safety....
High-impact/high-risk systems
Systems likely to affect rights, health, safety, economic inclusion, or security....
High‑Risk Artificial Intelligence
AI systems materially affecting life, safety, or fundamental rights....
High‑impact (high‑risk) system
Systems with substantial effects on safety, health or fundamental rights....