Canada and AIDA: the federal AI approach after Bill C-27 stalled

Short answer: Canada tried to enact a federal AI law, the Artificial Intelligence and Data Act (AIDA), as part of Bill C-27. When the federal Parliament was prorogued in early 2025, the entire bill lapsed. Canada therefore currently has no binding federal AI law like the EU AI Act; what remains is a voluntary code of conduct plus existing privacy, competition and liability law.

What AIDA would have governed

AIDA was meant to regulate "high-impact" AI systems: those posing significant risk to health, safety or fundamental rights. The proposal required managers and providers to assess, mitigate and document risks, be transparent about use, and accept a new regulator (an AI and Data Commissioner). The concrete meaning of "high-impact" was left to later implementing regulations — and a key criticism was precisely that the act deferred too much to future rulemaking.

Why it stalled

Bill C-27 bundled three things: a new privacy act, a tribunal and AIDA. The bill sat in committee for years. In January 2025 the government prorogued Parliament, causing all unpassed legislation, including C-27, to lapse. A new government would have to start over. There is no enacted successor at present.

What applies now

Lacking a statute, Canada relies on a voluntary code of conduct for advanced generative AI systems, to which leading firms have committed. Existing rules also apply: federal and provincial privacy law, consumer protection, competition law and liability law. The difference from the EU is fundamental: where the AI Act imposes binding obligations up front, Canada relies for now on self-regulation and generic law.

Place on the world map

Canada's path illustrates how fragile legislative momentum is. The country was early with a national AI strategy and a strong research ecosystem, but stalled on the political route to binding rules — unlike South Korea, which did get its AI Basic Act through parliament. Compared with Singapore, which deliberately favours voluntary frameworks, Canada has ended up in a similar position by necessity: governance without a binding AI law. See also our comparison of international AI governance.

What this means

• Do not treat AIDA as a compliance anchor: the proposal is not in force and may return in amended form.
• Follow existing law: privacy, consumer and liability rules are Canada's binding layer for now.
• Build on the shared grammar: governance based on the EU's risk-based structure and the NIST framework leaves you prepared should Canada later adopt a binding law.