As the United States pushes a lighter-touch approach to AI regulation and the European Union enforces its binding AI Act, Canada's own position sits somewhere in between. The federal government has leaned on voluntary codes of conduct and sector-specific guidance rather than a single comprehensive AI law, a strategy critics say leaves gaps as adoption accelerates across finance, health care and public services, sectors where the consequences of an ungoverned AI failure fall directly on consumers and patients rather than on the companies deploying the systems.
Supporters of Ottawa's approach argue it lets Canadian companies move faster than EU competitors while still signalling alignment with the international safety norms being pushed by the UN and allied governments, rather than either committing to the EU's compliance burden or appearing indifferent to AI risk the way critics accuse Washington's approach of being. The bet is that voluntary codes, if genuinely adopted at scale, can deliver most of the safety benefit of binding regulation without the enforcement costs and innovation drag critics attribute to the EU model.
Whether that bet pays off depends heavily on adoption rates that are difficult to verify from outside government, since voluntary codes by definition don't carry the same public compliance reporting that binding law requires. The debate is set to sharpen further as Canada's own AI infrastructure commitments deepen: Saskatchewan's sovereign compute project and the country's new national AI literacy programme both raise the stakes of getting governance right, since a major AI failure tied to infrastructure Ottawa itself helped fund would be a far harder story for the government to distance itself from than a failure at a purely private company operating under voluntary guidelines.
