Artificial general intelligence (AGI) describes a hypothetical system able to perform any intellectual task a person can, across arbitrary domains — not a system built and evaluated against one job. No fielded system meets that bar, and no Canadian statute, regulator or standards body defines or tests for it.
Every reachable Canadian government source on AI stops at describing capability within a bounded context, never at claiming generality. the Cyber Centre’s frontier-AI guidance defines “frontier AI” — the closest a Canadian government publication comes to describing today’s most capable systems — as encompassing “the most recent, capable and advanced models available”, and states that “These models have accelerated capabilities and have been shown to exceed the performance expectations in the field of cyber security”. That is a claim about performance in one field, cyber security, not a claim about general intelligence — the Cyber Centre’s own guidance never uses the term “AGI.”
the Canadian Artificial Intelligence Safety Institute, Canada’s own federal AI-safety body, works the same way. Its stated purpose is to “advance the science of AI safety, in collaboration with international partners”, and the risks it names are specific, not general: “synthetic content, including impersonation and fraud, as well as risks posed by the development or deployment of systems that may be dangerous or hinder human oversight.” Even the body Canada created specifically to watch for the most concerning AI risks catalogues named hazards rather than asserting that any system has become generally intelligent.
A vendor pitching a Canadian business on an “AGI-powered” product is making a claim about scope, not just capability. The useful question is the one implicit in NIST’s own definition of an ordinary AI system — that it operates “for a given set of objectives” — namely: what specific objective was this system actually tested against, and on what data? No Canadian regulator has published a test for “general” capability, so a claim of it cannot currently be verified against anything an official source has defined.
See also narrow AI, artificial intelligence and AI model.
This is one term in a plain-English glossary on how AI actually works and where it fits in a Canadian business.
