The End of a Comprehensive AI Framework
On June 4, 2026, Prime Minister Mark Carney unveiled Canada's "AI for All" national AI strategy — a five-pillar, $2.3 billion commitment to AI adoption, sovereign compute infrastructure, and tech-sector scaling. What the strategy conspicuously omits is equally significant: any plan to revive Canada's Artificial Intelligence and Data Act (AIDA), the comprehensive AI regulatory framework that died alongside Bill C-27 when Parliament prorogued in January 2025.
The case for comprehensive AI regulation was always serious. Proponents of AIDA argued that a horizontal framework would create predictable rules for high-risk AI systems across all sectors, rather than leaving enforcement to a patchwork of sector-specific regulators. The concern that powerful AI systems might cause harm — in financial decisions, medical diagnostics, or immigration adjudication — without dedicated legal accountability is not trivial. The EU's AI Act offers the counterexample: a single rulebook for an entire economic bloc, with industry knowing exactly what a "high-risk system" designation entails.
But AIDA's collapse was not merely political misfortune. The bill spent over four years in Parliament's pipeline — introduced as Bill C-11 in November 2020, reintroduced as Bill C-27 in June 2022, referred to the Standing Committee on Industry and Technology in April 2023, and finally killed when Parliament was prorogued on January 6, 2025, following Prime Minister Justin Trudeau's resignation. Committee hearings across 31 sessions and 69 witnesses produced an extensive legislative record, but no statute. Canada now relies on PIPEDA — a privacy law enacted in 2000, predating modern cloud computing — as its primary data and AI governance instrument.
What 'AI for All' Actually Does
The Carney government's answer is not a replacement omnibus framework but a deliberate fragmentation strategy: invest heavily in the sector, regulate narrowly where harm is acute, and let sectoral regulators fill the gaps.
The headline commitment is a $500 million Canadian Tech Growth Fund, which authorises the federal government to take equity stakes in promising AI companies. The logic is explicitly industrial: the fund targets firms that risk being acquired abroad or losing talent to larger U.S. competitors. Rather than imposing obligations on AI developers, Ottawa is positioning itself as a co-investor in their success.
The strategy also expands the AI Compute Access Fund to $1 billion for public supercomputing infrastructure — a recognition that compute access is now a competitive moat — and commits $200 million to an AI Missions Program focused initially on healthcare applications. Total announced commitments across the strategy exceed $2.3 billion.
The regulatory piece is deliberately narrow. Rather than waiting for an AIDA successor, the government has pushed Bill C-16 (Protecting Victims Act), introduced in December 2025, which criminalises non-consensual distribution of AI-generated intimate images. A May 11, 2026 amendment by the House of Commons Justice Committee expanded the definition to include "nearly nude" AI-generated images and imposed a 48-hour platform takedown deadline after notice. High-risk AI systems in finance, employment, or criminal justice get no equivalent dedicated statute.
The Upside — and the Gap
From a pro-innovation perspective, this approach has genuine advantages. Omnibus horizontal AI regulation carries a well-documented risk of regulatory capture and definitional overreach; the EU AI Act's classification system is still being contested by industry and member-state regulators. A targeted harm-specific approach — criminalise deepfake sexual abuse, invest in safe AI compute, develop sector-specific guidance — avoids locking in rules during a period of rapid technological change.
The $500M Tech Growth Fund is also conceptually sound. Canada has repeatedly seen promising AI companies — many with roots in the University of Toronto and Montréal ecosystems — absorbed by U.S. acquirers. An equity stake mechanism gives government financial upside while providing firms with growth capital that does not require a U.S. exit. This is the kind of intervention that is proportionate and reversible.
"Canada's approach avoids the EU's trap of premature rule-setting — but it also avoids the EU's clarity."
The gap is accountability. Business AI adoption stands at just 12% today; the government's own strategy targets 60% by 2034, implying a fivefold increase over eight years. Scaling adoption without a framework for accountability creates real exposure: when an AI system in a Canadian bank, hospital, or immigration tribunal causes harm, the legal recourse under current law is contested. PIPEDA was designed for human data handlers; its application to algorithmic decision-making was never settled even during AIDA's four-year committee review.
A Bet on Pragmatism
The Carney government's calculation appears to be that the cost of no AI law is lower than the cost of a bad one passed hastily. Given AIDA's track record — four years of parliamentary effort, zero statutes — that is not an unreasonable reading of Canadian legislative politics. Parliament's record on digital legislation is one of ambitious introduction and repeated attrition.
Whether the bet holds depends on the next several years. If AI adoption scales toward the government's targets and harms remain manageable through existing sectoral oversight, the fragmented approach will look prescient. If a high-profile AI failure occurs in an unregulated sector, the absence of AIDA will become politically untenable.
What Ottawa has done is make a deliberate choice: growth capital and targeted harm prevention now, comprehensive governance later — if ever. That choice deserves scrutiny, not because comprehensive regulation is obviously correct, but because the regulatory vacuum it leaves is not neutral. Canada's 12% AI adoption rate is both a problem the strategy addresses and a warning: a country moving quickly toward 60% adoption on an accountability framework written before the iPhone will eventually face a reckoning. The question is whether it plans for it, or waits for the incident.