Last updated on June 26th, 2026 at 12:53 am

Part of the Canadian AI Sovereignty Series

The dominant narrative for years about Canada’s digital strategy has been that the Trudeau position was embarrassing. How dare we expect US giants like Meta and Google to pay a fee, essentially, to access Canadian audiences. That sentiment came from industry, particularly the technology industry, and it is the one Michael Geist has echoed again in his latest take on the digital policy cavalcade of the past few weeks. (This cavalcade brought on a proposed legislation to reverse the streaming act, a long-delayed AI strategy, a social media ban for under-16s paired with a new superregulator for platforms and AI chatbots, a privacy bill that strips the Privacy Commissioner of authority over private-sector privacy, and lawful access legislation rushed through committee.) We can say yes, much of this is bad legislation, but let’s not obscure where much of the criticism comes from, two different things.

Geist is once again criticizing regulation of the industry, and we hear it again in his op-ed today in the Globe and Mail. This is the default position that many have now accepted as the way things should operate in Canada. People ask why we cannot see news on Facebook and answer themselves with “the government,” when in fact the government took a principled position about why access to Canadian audiences should be funded by the giants.

Here is how that actually works, and how it all transpired.

The law was Bill C-18, the Online News Act, introduced in 2022. It demanded that Google and Meta pay Canadian publishers for distributing their content, on the straightforward premise that two companies holding a combined eighty per cent of a fourteen-billion-dollar online advertising market, in a period that saw hundreds of Canadian news outlets close, were drawing value from journalism they did not fund. The bill passed in June 2023. Both companies threatened to block Canadian news links rather than pay. By September the government had put numbers to its expectations: roughly $172 million a year from Google, $62 million from Meta.

Then the two companies split, and the reason they split is the whole story. Google needed the news. Not for ad revenue, but because a search engine that cannot return news is a worse search engine, and Google was willing to pay to keep its results comprehensive. So Google came to the table. On November 29, 2023, it agreed to pay around $100 million a year, indexed to inflation, into a fund distributed to eligible Canadian news organizations. News still lives on Google in Canada today because Google decided the access was worth funding.

Meta did the opposite arithmetic. Its revenue does not depend on news the way Google’s does, since links are posted by users rather than pulled and curated by the platform. Meta concluded it needed Canadian news less than Canadian news needed it, and it walked. It blocked news on Facebook and Instagram, and the block has now stood for more than two and a half years. Canadians still cannot see news on either platform.

The damage from that walk is real. Within months, Canadian news views on Facebook had fallen by roughly five million per day, and the smaller and newer outlets that depended most on Facebook for distribution were hit hardest, some losing more than half their audience. Meta itself barely felt it. Canadian usage of its platforms stayed flat, and observers noted the company came away emboldened, having proven it could simply leave a national news market with no consequence to its bottom line.

And who lined up where? Faced with a government saying the platforms monetizing Canadian attention should pay something back into the ecosystem they draw from, much of Canadian industry, and a good deal of Canadian commentary, sided with the platform that left. The framing settled into place quickly: the law was unworkable, the government was naive, the giants could not reasonably be expected to pay. The visible result of that position is a news sector that lost its Facebook distribution entirely and got, from the one company that stayed, a fund many recipients now have reason to feel ambivalent about. Most of the discourse read this episode as Canada overreaching. What it shows is one platform that found access worth paying for and one that found it could take Canadian audiences for free, and a country that let it.

Donald Trump keeps floating the idea of Canada as a fifty-first state, which we all agree is ridiculous, but commercially the thing he is describing has in many ways already happened. The point is not that Canada should build its own Netflix (though to a degree we have, in Crave which produced global smash Heated Rivalry but is itself a service heavily reliant on American programming) but that there has to be a balance.

This is the crux of Canadian identity, the thing so many Canadian commentators are fond of describing as weakened by multiculturalism. It is not being weakened by multiculturalism. It is being weakened by a lack of funding, in favour of acceding to American corporate interests. Our news industry has been decimated in plain view. Canada’s largest newspaper chain, Postmedia, more than a hundred and thirty titles including the National Post and most of the country’s metro dailies, is majority-owned by Chatham Asset Management, a New Jersey hedge fund with Republican ties, with a further chunk held by a second American hedge-fund investor. That chain draws Canadian subsidy and a share of the Google news money while routing tens of millions to its American owners every year through high-interest debt. Why would we not regulate this? How are the arguments against doing so still treated as valid? The cavalcade is not the best legislation we have ever produced, and there is a great deal in it I do not agree with. There is overreach in it. There is also underreach, and we cannot seem to solve either problem effectively. What we should be doing is protecting our own industries by charging for access to our attention and protecting the privacy of Canadians. Right now we are doing neither.

A risk model tuned to one failure

The instinct underneath the critique was built for a world that no longer operates the way it did.

Geist’s analysis optimizes against a single failure mode, and that failure is overreach. The state reaches too far, and the costs that follow are the costs he tracks: weakened security, captured markets, regulatory burden, the government substituting its judgment for the market’s. For most of the period he has been writing, watching that failure was the right call. The downside risks of Canadian digital policy were mostly domestic and mostly reversible. A badly built streaming rule can be unwound. An overzealous regulator can be reined in. When the dominant cost is the state doing too much, a framework that scores every intervention against the overreach baseline will usually point the right way. That baseline has expired.

When the exposures a country faces are foreign jurisdictional reach into its data, dependency on systems it cannot audit or revoke, and infrastructure owned and governed elsewhere, the structure flips. The danger is no longer the state acting too aggressively. It is the state failing to act, or acting too late, on capacity it does not control. A risk model tuned entirely to overreach cannot register that, because inside that model state action is the thing being guarded against rather than a tool that might be applied too sparingly. So weeks of restrictive, centralizing, clumsily built policy get scored the way such weeks have always been scored. The verdict is not wrong about the clumsiness. It is measuring against a world where clumsiness by the state is the primary harm, and that world is gone.

The jurisdiction the column skips

The lawful access legislation is where this shows most clearly, and it is also where Geist’s own column has a revealing gap. He reads the bill as a domestic privacy and security story: mandated metadata retention, weakened encryption, the alarm of security researchers, the risk that essential security firms leave the country. Every one of those concerns is real.
The CLOUD Act does not appear anywhere in the analysis.

That is a large absence in a piece about compelled data retention. Mandated metadata retention manufactures the dataset. Foreign jurisdiction reaches it. If the providers holding that retained data are American-headquartered or American-owned, Canada has legislated a richer trove into existence and left it sitting under the reach of a foreign government’s legal process. The exodus of security firms Geist flags is the second-order worry. The first-order one is that the firms which stay are precisely the ones already exposed to foreign compulsion. Domestic retention and extraterritorial access are the same vulnerability seen from two ends.

The governance frame buries this. If the problem is Canadian governance architecture, the wrong minister holding the privacy file, a regulator that moves too slowly, then the solution space is entirely domestic. Restructure the org chart. Centralize the decisions. Stand up the superregulator. None of it touches who can compel the data once the data exists. A country can fix every governance complaint in the column and still have built a surveillance dataset reachable by a foreign state. The jurisdictional layer sits at a right angle to the org-chart layer, and folding everything into governance makes the more serious exposure vanish.

What happens when a vendor decides

The exposure is not only national. It is personal, and most Canadians have already met it.
A person loses access to a Twitter or X account they have held for fifteen years, with the followers, the archive, and the working relationships attached to it, and there is no appeal that resolves anything, no regulator to call, no obligation on the company to explain or restore. Someone buys a film on Amazon, watches it, returns months later to find it gone from their library because a licensing deal lapsed somewhere they will never see, and the purchase they thought they made turns out to have been a revocable permission all along. A small business builds its customer reach on an account that a platform suspends without notice, and the platform owes it nothing. These are not edge cases. They are the ordinary terms of the relationship, and the terms run entirely one way.

There is no recourse in any of it, and the absence of recourse is the point. The platform sets the rules, changes the rules, and enforces the rules, and the user agreed to all of it in a contract no one reads and no one could negotiate. Calling it naive to want a floor under that, a baseline of notice, appeal, and durability of the things people have paid for or built, is not realism. It is consumer protection, the most ordinary kind, the kind Canadians take for granted from their bank and their utility and their phone company although even that is degrading now. The reason it has not arrived for digital platforms is not that it would be naive but because the platforms have spent a decade telling Canadians it would be, and a great deal of Canadian commentary has agreed.

That is the question the overreach framework cannot answer. If a company can erase what you own and silence what you have built with no obligation to you whatsoever, the danger in the relationship is not the state stepping in. It is the state continuing to stay out. Why a country would keep caving to that, keep accepting that an asymmetry this total is simply how things work, is the thing no one defending the platforms’ position ever has to explain.

What the Netflix deal actually was

The era Geist files under naive cheerleading, and the framework’s blind spot, becomes a pattern rather than a single miss.

His version 1.0 is the hands-off, technology-embracing phase, and the emblem of it is the five-year, five-hundred-million-dollar commitment Mélanie Joly secured from Netflix in 2017. He codes that as a government too eager to demonstrate openness, charmed by tech, light on regulation.
The deal was attacked at the time from the opposite direction. Quebec’s culture minister called it Netflix paying itself half a billion dollars to make content it would sell for profit, an unfair edge over Canadian companies. Analysts noted the annual figure was barely one and a half per cent of Netflix’s spending, and that the company still owed no sales tax, no corporate tax, and no content quota. The same establishment that now files the deal under naive overreach attacked it then as a corporate giveaway too soft on the platform. It cannot have been both. A deal that gets read as whatever indicts the government in the moment is not being analyzed. The conclusion is fixed and the reasoning runs backward from it.

Look at what the deal was. It was a condition of market access. You want the Canadian audience, you fund Canadian production. That is not a government failing to understand technology. That is a government correctly identifying that access to its market is the leverage it holds, and pricing it. The vendors were always going to extract value from Canadian attention. The only live question was whether Canada extracted anything in return.

The same structure runs through the streaming act and the news bargaining that followed. Geist scores those as botched. Three separate things were bundled under “botched,” and the third is vendor-driven. The distinction matters, because execution failure vs vendor obstruction discredit the principle.

First, the drafting choice. The government wrote C-11 as a broad enabling statute. It handed the CRTC wide authority and vague definitions and left the hard substance, what counts as Canadian content, how much streamers pay, what “discoverability” means, to be worked out later through CRTC proceedings. That was Parliament’s decision, not a vendor’s. It’s the same flaw Geist names in C-18: too little detail in the law, too much left downstream. So the bill was structurally slow before any platform did anything.

Second, the regulator’s pace. The CRTC pushed implementation from late 2023 to late 2024 to late 2025, with consultations still running into 2026. Industry experts greeted the delay with dismay, noting the CRTC had told Parliament in May 2023 it would implement the legislation by the end of that year.  The chair’s defense is that the work means completely overhauling decades-old regulatory frameworks, with many separate and tricky issues, including very different English and French markets.  Real complexity, but also a regulator moving at administrative-tribunal speed on something the sector needed fast. Again, not vendor sabotage, just slowness, which is the thing Marc Miller publicly attacked the CRTC for before the June reversal.

Third is the part that is vendor obstruction, the litigation. In July 2024 the Motion Picture Association–Canada and the Digital Media Association sued, objecting to the requirement that streaming services contribute to local news, with MPA-C calling it regulatory overreach and a discriminatory measure beyond what Parliament intended. The principle underneath, that a platform monetizing Canadian attention owes something back to the ecosystem it is drawing from, is not policy weakness. It is the recognition that the overreach in the relationship belongs to the vendor. Google and Meta built businesses on Canadian news and Canadian audiences and routed the value out of the country. Requiring them to fund the access is not the state overreaching into a free market. It is the state belatedly noticing the market was never free, it was captured, and the law was an attempt to price the capture.

This is the same inversion as the lawful access point, which is why the two belong in one argument. The framework can see overreach running in only one direction, from state onto vendor. It has no slot for overreach running from vendor onto country. So a funding obligation reads as a shakedown, foreign jurisdictional reach reads as a footnote, and a half-billion-dollar production commitment reads as a starstruck mistake. Reverse the direction the overreach travels and every one of those re-sorts. The Netflix deal becomes Canada doing the thing correctly. A news sector on life support becomes the picture of what happens when the country stops.

The choice that is actually on the table

The Trudeau government took the funding approach, and the industry it was trying to defend handed it a great deal of criticism for the attempt. That criticism was largely unearned, and the alternative it implies is is significant.

The choice is not between clumsy Canadian regulation and some frictionless free market. It is between requiring the platforms that monetize this country to fund a share of what they draw from, and telling them the country is theirs for the taking. Open the doors to the American firms, accept whatever falls from the table, and call the absence of obligation a principle. That is the policy hiding inside the critique. Demanding the platforms pay into the ecosystem is the thing that protects a Canadian news industry currently kept alive on the thinnest of margins.
The stakes there are not abstract. A news sector starved of funding is itself a sovereignty exposure. It is the civic layer meant to be watching all of this, the procurement decisions, the dependencies, the two weeks in June, and it goes dark for want of the exact funding mechanism the overreach framework spent a decade opposing. The institution that would have scrutinized digital strategy 3.0 is the institution that framework helped defund.

None of this requires pretending the laws were well built. They frequently were not. The Online News Act’s rollout was clumsy and the link-blocking did real harm. But “the law was clumsy” and “making platforms fund their access is wrong in principle” are two different claims, and the overreach framework consistently lets the first quietly discredit the second. Pull them apart and the case against the principle thins out fast.

We live in a different world than the one this framework was built to navigate. The exposures are larger, less reversible, and increasingly governed from outside the country. A risk model that treats every state intervention as the thing to fear was the right instrument for an earlier moment. Pointed at this one, it keeps returning a confident reading of the wrong threat, and it cannot see the overreach that is actually doing the damage, because that overreach is running the other way.