Canada’s AI strategy: Clear direction or digital self-sabotage?

Log in to listen to this article

Canada’s much-anticipated national AI strategy was released last month to mixed reviews.

The federal government's plan, AI for All, aims to drive AI adoption among Canadians through six pillars: safeguarding Canadian democracy, AI literacy, supporting AI adoption in small and medium-sized businesses, Canadian tech sovereignty, economic development, and innovation. 

Experts, however, are skeptical that the strategy will foster public trust in the technology or keep tech companies accountable.

“Instead of regulating AI, they’re regulating digital spaces,” says Brent Arnold, chair of the Canadian Internet Society and partner at INQ Law. 

“You have to have a clear strategy to regulate AI. We’re trying to do a lot of things at once, but I don’t think the entire plan will be 100 per cent successful.”

The focus is on economic development, and much of the plan includes items on job creation, pushing for the majority of small and medium-sized Canadian businesses to adopt AI by 2034, and infrastructure investments, including building more data centres.

"The strategy provides good direction,” says Ashley Casovan, managing director of the International Association of Privacy Professionals’ AI Governance Center. The IAPP was among the hundreds of organizations that participated in the feds’ short consultation process last year.

“In particular, investing in the pipeline of talent … and recognizing that we need to prioritize different areas, as well as continue to build out strategic global partnerships, is important. The emphasis on having people at the core to support AI adoption is critical.”

Other experts have less praise for the plan. In a piece published in the Globe and Mail, Michael Geist, the Canada Research Chair in Internet and E-Commerce Law at the University of Ottawa's Faculty of Law, said the strategy suffers from “digital self-sabotage.” 

“Canada cannot become an AI leader while it builds a reputation for poorly conceived policies that threaten security, make internet services uneconomic and raise new barriers to the very adoption the strategy is meant to encourage,” he wrote. 

No new legislation

The strategy mentions legislative reforms, such as the newly announced Safe Social Media Act, but doesn’t commit to creating any new AI-related legislation. This is a new shift away from creating a single law to regulate AI. In 2024, the European Union enacted the Artificial Intelligence Act, which requires organizations to conduct risk assessments when implementing AI, respect copyright law when training AI systems, and ensure human oversight in high-risk activities. 

Canada initially appeared set to follow the EU model. In 2022, the federal government tabled the Artificial Intelligence and Data Act (AIDA), which was similar to the European legislation. The bill died when Parliament was prorogued in 2025. 

Under the Biden administration, there was momentum in the United States to create AI legislation. That changed with the Trump administration’s executive order in 2025, where U.S. dominance in AI became the priority, and any mention of developing safety regulations was removed. 

“It’s clear that the (Canadian) government has no intention to introduce a broad AI risk-regulation type statute,” says Teresa Scassa, a law professor at the University of Ottawa and the Canada Research Chair in Information Law and Policy. 

“The minister of AI and digital technology has talked about AI regulation that is ’light, tight and right.’ The government has achieved ’light.’ There doesn’t seem to be much of a plan for ‘’tight or ‘right.’”

The problem is that the new federal strategy creates a mix of AI policies and regulations that will make it difficult for businesses to comply with the law. It’s a similar situation in the U.S., where several states have created AI and privacy laws because there are no U.S. federal laws regulating the technology.  

“To comply with the new laws, you’re going to have to grapple with different conflicting laws like creating safety assessments for different platforms,” says Arnold. 

“Regulating AI through a series of laws dealing with adjacent issues won’t work.”

Despite the new federal strategy, Arnold disagrees that Canada has abandoned plans for AI legislation.

“The main complaint of AIDA was that it left almost everything in the regulations,” he says. 

“What does that mean for companies? We couldn’t say whether a company was compliant until the regulations were written. That means companies didn’t know what to expect, creating uncertainty. We might still have another attempt for AI legislation in the future.”

One major barrier to AI legislation is the growing power of tech companies. However, Arnold says the divide-and-conquer strategy with tech companies threatening to pull services is working in the face of growing pushback. Florida's attorney general has filed a lawsuit against OpenAI for endangering and addicting children, coaxing users into suicide, and aiding and abetting mass shooters. The family of a victim of a 2025 mass shooting at Florida State University is also suing the company. The suit alleges ChatGPT enabled the attack, as the shooter carried out the attack using advice and instructions the technology generated.

Victims of the Tumbler Ridge shooting have filed similar lawsuits. Their families allege that OpenAI could and should have prevented the shooting earlier this year. 

There are also growing protests against the construction of data centres. Several U.S. cities have blocked or stalled data centre construction projects. 

“We have to continue to try to regulate AI,” says Arnold. 

“(Companies will leave) is not a good excuse to give up on regulation.”

Adoption requires literacy

One of the positive stories emerging from Canada’s AI strategy is its emphasis on AI literacy. The plan calls for creating educational programs for postsecondary students and providing funding to community organizations and libraries to support AI programs for adults. Scassa says increased literacy is needed to support any efforts to increase AI adoption. 

“For businesses, it is hard to adopt AI if they are not clear what the benefits will be or how they should adopt/integrate it most effectively,” she says. 

“There are obvious risks to doing it in the wrong way. There may also be issues around having the capacity to adopt and integrate AI, which may relate to skills training. AI literacy is important.” 

Encouraging AI adoption is complex, however. There are growing questions about the environmental impact of AI use, given the incredible water use and carbon emissions of data centres. There are also ongoing issues around incorporating Indigenous, First Nations and Métis data governance principles into regulations. The federal strategy points to the need for more consultation but doesn’t list concrete steps.

“There’s a bargain going on where the federal government is saying let’s pursue economic development in AI, and some things will be sacrificed like the environment,” Arnold says. 

“There needs to be a larger discussion about whether to be competitive, we can’t have regulation.” 

He doesn’t believe that’s the case, that it’s a matter of either or.

“We could see what was coming. (First), general large language models were here, and now more purpose-built tools are being developed. We got distracted, and now it’s here.” 

For lawyers, a complex regulatory system will be another challenge in adopting AI. Lawyers will need to understand how the new policies and regulations intersect with conflicting laws in the U.S. and the EU. Cases involving AI hallucinations continue to make headlines, despite all law societies in Canada now having AI regulatory guidance. As lawyers continue to figure out how AI fits in their legal practice, they will continue to balance advising their clients on AI regulation. Scassa says, 

“Right now, AI law is complex, multi-faceted, and changing and evolving all the time.”