Eighteen months after the Artificial Intelligence and Data Act died with Bill C-27 on the order paper, Canada's innovation ministry has opened the first concrete regulatory follow-through of the post-AIDA era — and it is a consultation, not a bill.
Innovation, Science and Economic Development Canada (ISED) launched "Have your say on advancing AI transparency in Canada" on July 23, 2026, running through September 23. The discussion paper, Enhancing Trust in Artificial Intelligence Through Increased Transparency, asks Canadians and industry to weigh in on five areas: labeling AI-generated content, disclosing when someone is interacting with an AI system rather than a human, publishing information about what AI systems can and cannot do, tracking serious AI-related incidents, and monitoring the behavior of autonomous AI agents.
The steelman: adoption is outrunning the guardrails
The timing case is real. Statistics Canada's most recent survey found that 19.2% of Canadian businesses reported using AI to produce goods or deliver services in the 12 months to Q2 2026 — up from 12.2% a year earlier, roughly tripling in two years. Chatbots and virtual agents are now among the most common deployments. When adoption moves that fast, the argument for at least some baseline disclosure — so a consumer knows whether they're arguing with software before it makes them a binding offer — is not manufactured urgency. It reflects a genuine gap: Canada has no general AI statute, and sector regulators (privacy, competition, broadcasting) each see only their slice of the problem.
The consultation also sits inside a broader strategy with real political weight behind it. Prime Minister Mark Carney launched "AI for All" on June 4, 2026, a five-year plan targeting $200 billion in additional GDP and a jump from roughly 12% to 60% AI adoption by 2034, built on three pillars including "protecting Canadians and safeguarding democracy." Transparency is the connective tissue: a strategy premised on public trust cannot simultaneously ignore the question of whether people know when they're dealing with a machine.
Why soft law is the right sequencing, not a retreat
The more interesting signal is what the discussion paper does not propose. Unlike AIDA's risk-tiered, high-impact-systems framework — which drew years of criticism for defining "high-impact" too vaguely and layering criminal offences onto a fast-moving technical domain — this consultation explicitly asks respondents to weigh a mixed toolkit: legislation, the existing Voluntary Code of Conduct on generative AI, Standards Council of Canada guidance, and Canadian AI Safety Institute investment, rather than assuming a single binding statute is the answer. It directly asks whether "existing market responses and legal frameworks are already adequate."
That is the correct order of operations. AIDA's fatal flaw wasn't that it addressed transparency — it's that it tried to codify enforcement mechanisms for a technology whose deployment patterns hadn't stabilized, and it died without a single hour of parliamentary debate on its substance. A consultation that surveys which disclosures actually change user behavior, and which are box-ticking, is the homework Canada skipped the first time.
That said, two of this consultation's five buckets — AI-agent monitoring and incident tracking — are genuinely harder than the EU's approach, which the Canadian paper is clearly aware sits nearby. The EU AI Act's Article 50 transparency obligations — machine-readable marking of synthetic content, disclosure of AI interaction, and deepfake labeling — took effect August 2, 2026, three weeks before Canada's consultation opened, with fines up to €15 million or 3% of global turnover for noncompliance. Ottawa now has a live natural experiment next door: whether Brussels' binding, penalty-backed labeling regime actually reduces synthetic-media harm or mostly generates compliance paperwork for developers who were never the problem. Waiting even six months to see early EU enforcement patterns costs Canada nothing and could save it from importing a rule that turns out to be more theater than protection.
The parallel track that will actually bind sooner
Canadians should not mistake this consultation for the whole picture. ISED is simultaneously advancing Bill C-34, the Safe Social Media Act, introduced June 10, 2026, which explicitly covers AI chatbot services and would create a Digital Safety Commission with binding obligations — a harder-edged companion to the soft-law transparency consultation. If C-34 passes first, some of what this consultation is polling on (interaction disclosure for chatbots, at minimum) will already be settled by statute before ISED even publishes its "What We Heard" report.
The honest read: this consultation is not Canada's AI law. It's the input-gathering phase that AIDA never properly completed, running in parallel with a narrower, faster-moving bill that may pre-empt part of its own subject matter. For a government still recovering credibility after AIDA's collapse, asking the question before legislating the answer is the more defensible sequence — provided the "What We Heard" report actually shapes what comes next, rather than serving as a consultation-shaped rubber stamp for a transparency bill already drafted in a ministerial drawer.