Sabotage by screenpunk, https://flic.kr/p/4FZMPs CC BY-NC 2.0

The Canadian government’s long-awaited and much-needed AI strategy is finally set to be unveiled this week, with AI minister Evan Solomon promising a plan that prioritizes AI adoption, investment, and regulatory guardrails to enhance trust, privacy and safety. My Globe and Mail op-ed argues the strategy seems doomed to fail, even before it is released, with the government’s own digital policies working against it. An astonishing series of developments in recent weeks amount to digital self-sabotage, leaving global technology giants alarmed and Canadian tech companies openly considering leaving the country.

The contradiction is hard to ignore. 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.

Start with security and Bill C-22, the much-criticized lawful-access legislation now before a House of Commons committee. It has prompted near-daily announcements by companies prepared to leave Canada due to the risks posed by encryption and privacy-invasive mandatory metadata retention requirements. Signal has said it would exit the Canadian market rather than comply; virtual-private-network services such as Windscribe and NordVPN have said the same, and search engine DuckDuckGo has said it would curtail services here. The biggest tech giants, such as Apple, Meta, and Google, have all descended on Ottawa to urge changes before it is too late.

These are not idle threats, as there is a history of companies exiting countries that follow through with similar legislation. And the timing could hardly be worse, given that Anthropic’s Claude Mythos model has shown that AI can now find and exploit security vulnerabilities on its own, including weaknesses in the encryption currently in place to guard private communications. Introducing new security vulnerabilities into the system sends a clear message that working in Canada presents real risks for technology and AI services.

If security is the fear with Bill C-22, economics are the problem with the government’s approach to Internet streaming services. Last month, the CRTC tripled the mandated contributions and expenditures for foreign streaming services to 15 per cent of their Canadian revenue, among the highest amounts anywhere in the world. Moreover, new discoverability rules not only regulate streamers but also govern how content is presented on smart TVs and set-top boxes, and even which apps are installed on the devices.

There are serious concerns that the mandated payments, which require streamers to partner with Canadian firms on film and television productions but preclude them from owning the resulting intellectual property, violate Canada’s trade obligations. This is particularly bad timing, as we are set to undergo a review of the U.S.-Mexico-Canada Trade Agreement. Ottawa is sending a clear message to the tech sector: investing in the Canadian market means dealing with a steady stream of trade and legal battles.

Meanwhile, the AI strategy itself threatens to create more barriers to AI adoption. The strategy is expected to include mandated age verification for using both social media and AI chatbots, with a plan to establish new restrictions on the use of these services for those under the age of 16. But limiting AI access to adults will require everyone to certify their age, typically through foreign third-party services that introduce new privacy risks to sensitive personal information.

You can’t proclaim AI leadership while eschewing rules that address safety issues for all users by instead relying on age-gating services with requirements that will affect tens of millions of Canadians. Yet that is what the federal government is proposing to do with promises of “AI for all” that may actually become “ID for all”.

None of this is hypothetical, either, because Ottawa has been here before. When it passed the Online News Act, it brushed aside warnings that Meta would block news rather than pay for news links. And when Meta followed through on its threats – a blockade that is in place to this day – Canadians simply lost access to the links (and credible information) that the law was trying to protect. The lesson the government keeps ignoring is that when the rules become impossible or unpalatable to follow, companies are prepared to leave.

The forthcoming national strategy will surely seek to establish an ambitious vision for AI in Canada. But it cannot succeed while the government keeps advancing policies that make the country a riskier, costlier, and less welcoming place to build and use the technology.