Episode Transcript
EP. 35.1 "AI for All" of Carney's Friends
MATTHEW REMSKI: Hello everyone, this is Matthew Remski with Antifascist Dad Podcast, episode 35.1. This is a short bonus episode called "AI for All of Carney's Friends."
I'm grateful for your support. I hope this project brings some joy and hope and utility to your works and days. You can find me on Bluesky and Instagram under my name. I'm on YouTube and TikTok as antifascistdad, and the Patreon for this show — which, if you're listening to this episode on the day it drops, you know all about — is antifascistdadpodcast. Subscribers get early access to every second part of the main feed episodes, including this one. Sometimes it's an unrelated subject, so this is the follow-up episode to my interview with EV Debs, the game maker for Antifascist Tooth Fairy. Every single bonus episode gets released to the wild eventually. Really, the Patreon is about supporting this show and this work going forward.
A reminder that my book is now available. The link is in the show notes, and if you've got it — if you're reading it, if you're listening to the audiobook — please consider leaving a review wherever that makes sense.
Now, about that book. When I started it, the day after Trump's second electoral victory, I was only vaguely aware of who Mark Carney was. Bank of Canada, Bank of England, somber pressers during Brexit. I didn't pay that much mind. So I certainly didn't expect to be spending so much time and attention on this guy and on this moment and on this government. And yet, here we are.
For years, I was much more absorbed in US politics through my work on Conspirituality Podcast, and I was kind of sleeping on my own national beat, except for coverage of the Trucker Convoy and Danielle Smith's affinity for wellness pseudoscience. But now I cannot look away.
And while there is a danger in scope and concept creep with giving Carney so much attention, I do think that his efficacy as a proto-fascist leader is absolutely worth concentrating on in any intergenerational conversation about capitalism and fascism. Because it's just easy to see what Trump is up to. It's easy to call out the fascist dads. In fact, my first ten episodes here included a whole fascist dad segment at the end, but I dropped it because there are so many of them and it was getting repetitive. And also the constituency of the fascist dads, if they're locked in, will not be reachable through analysis for the most part.
But with liberal figures, I think it's different. They nurture the conditions by which capitalism comes to crisis, and they do it from within the affect and proceduralism of normalcy. They're the ones who can go from some scummy backroom deal to showing up at a middle school event and giving an inspiring speech. So in many ways, it's not the fascist dad I believe we have to concentrate on most. It's the liberal dad who makes the fascist dad welcome.
Now, with regard to the subject today, I think there's something about AI that is so generalized, so pervasive, so obviously on the nose with regard to how blatantly immoral it is. It's almost as if the workers of the world are being presented with the full implication of capitalism all at once. Wealthy people just openly saying: we have this wealth, see, and we're going to use it to build machines that will steal your data from you and then rent it back, while replacing your jobs. And also the machines will drink all your water and use all of your electricity and create heat domes over your towns. And then we'll demand big tax breaks from your councils besides. And if you complain, we'll shut you up. We'll sue you. We'll say you work for the Chinese Communists. We'll say you are bots.
Now, Carney doesn't get into threats and insults. He leaves that for other people. And that's what makes him the liberal dad, holding the door open for his fascist-friendly colleagues.
Speaking of which, let's get into Carney's AI strategy, released June 4 in the document "AI for All," which really crystallizes everything about his neoliberal content and his slippery form.
I want to take a look at some keywords. To start: the word "sovereignty" appears 51 times in this document. Tied for second in the keyword count at 35 hits were "data," which makes sense for the topic, and "trust," which I don't think does. Know what words are not in the document at all? Emissions, climate, pollution, unemployment, union, consent, surveillance — and then words like Palantir, Amazon, Microsoft, Google.
Now, first off, you should probably keep the word "sovereignty" out of your mouth unless you're a First Nations person using it to describe the freedom you don't yet have on treatied or unceded land. But unfortunately, "sovereignty" is one of those terms that aggrieved working-class folks — some of them racist — like to pick up and wave around like a flag, especially in Alberta, the epicentre of Canadian petrolibertarianism and now separatism.
But because a lot of these are deracinated folks with more connections to highways and talk radio than to the soil, they tend to use it to talk about body sovereignty. And that meant that during the height of the pandemic, they pretended that masks and vaccines were an assault on their free and immaculate bodies — on their bodily sovereignty — instead of being elements of a communitarian public health effort. Their media platforms — True North, Rebel, Western Standard — attached the term "sovereignty" to everything from gun rights to provincial jurisdiction to cryptocurrency, as in: you can be your own bank.
For the past several years, Conservative leader Pierre Poilievre, who rode hard with the Trucker Convoy, has been busy eliding the concept of bodily sovereignty with digital and data sovereignty, using the rhetoric of distrusting globalist control of Canadian data and speech. And on that last part, he's not exactly off base.
Today, Alberta Premier Danielle Smith is pushing something called the Sovereignty Act on behalf of her pseudo-nationalist supporters, on the off chance that it could hand over more money to the oil lobby that put her in power.
One caveat here: it would be a mistake to dismiss Alberta separatists as purely reactionary, because that would miss the fact that life in the Prairies is genuinely getting harder. The cost of living is rising. Decent work is drying up. The basic conditions that made the province's harsh climate bearable are eroding. Late capitalism is doing what it does, and ordinary people are feeling it in concrete ways. And the tragedy, as always, is the misdirection: too many Albertans are ideologically captured by the petro-economy that's actually organizing their immiseration, and they may not be able to see their way clear of it before the damage becomes irreversible.
Now, by the time Carney starts talking about "sovereign AI" — which is what this document is filled with — the word already carries this freight of anti-establishment, anti-globalist resonance that he gets to put in his pocket without earning it, and then set it against the demented ramblings of Trump.
However, "sovereign AI" is just a lie. Palantir holds pre-approved vendor status for government contracts. Three US firms own 85% of the Canadian cloud. And as I discussed in a recent Conspirituality episode, the 56-megawatt STACK data centre that's just two miles from my house hasn't even disclosed who its tenants are. If they were Canadian tenants, I'm sure that would be all over the advertising.
So Carney is using the word "sovereignty" to con public consent for a private hoarding of publicly subsidized infrastructure, facilitated by the state through a no-consult approvals process and total grid access.
Sovereignty means something to those who have been colonized, to those who live under neocolonialism. Then the colonizer working class lifts it because they're yearning for their own lost souls. And then the Davos class uses "sovereignty" to designate how they want to control global capital. It's just more evidence that for Carney, Canada is not a country — it's a brand.
So that's the top-line analysis. I'll add a bit more, especially in relation to what I said above about the sovereignty framing being undermined by the facts. Palantir holds pre-approved federal vendor status. Amazon, Microsoft, and Google control approximately 85% of Canada's public cloud market. And there are unnamed partnerships proposed to deliver 850 megawatts to 2.3 gigawatts of compute capacity by 2030. Nothing suggests these won't be drawn from the same foreign investor class that the strategy professes to reduce dependency on.
Another element: in November 2025, Carney signed a Canada-Alberta Memorandum of Understanding. This is a new sort of political instrument that's flying around these days, and I hope to interview Nora Loreto about it, because she's writing a book on what's been happening to Canadian democracy and how these federal-provincial agreements made at the individual level are functioning like agreements between independent nation states.
This MOU suspended the Clean Electricity Regulations and eliminated the Oil and Gas Emissions Cap, all in the wake of intense lobbying by the Capital Power gas company. The MOU also listed increasing oil production in Alberta as its first objective — this is the agreement between Carney and Smith — and proposed AI data centre development in Alberta as a new market for the province's gas.
In my last segment on this for Antifascist Dad, I noted that public opposition here in Canada was off to a slow start, but it is growing. Organized protests have emerged in Vancouver, with hundreds marching against the federally endorsed Telus data centres in May. In rural Alberta, the Sturgeon Lake Cree Nation has been vigorously opposing a Treaty 8 land transfer for a data centre project. And just west down the highway from me here in Toronto, hundreds of Hamilton residents showed up to a town committee meeting on June 4 to oppose a hyperscale plan for a decommissioned steel mill site — and they won an initial permit denial from the Committee of Adjustments. The corporation will fight back, of course. They might even sue the town. But this is a start.
Reading through this document is like reading through a business proposal or investor's deck or mission statement. It's all very aspirational. It's not about governance. The vibe is entrepreneurial. Every commitment uses "Canada will" rather than "our legislation will require" or "regulation mandates" or "there will be penalties for these violations." There is not a single bill number, no proposed regulatory body with new powers, no compliance deadlines, no enforcement mechanisms anywhere in the document.
These absences are most obvious where the aspiration runs thickest — all in the category of things that Carney's team knows will build trust, ostensibly by talking about them. Here is a list of the risks the document waves at but offers no concrete remedy for: consumer privacy modernization — nothing. Online safety laws — nothing. Watermarking or AI transparency standards — nothing. Protection of elections from AI disinformation — nothing. Regulation on algorithmic decision-making in hiring, lending, and healthcare — also nothing.
Now, who's out there trying to sell this plan to the public? Front and centre in the effort is Evan Solomon, who is a walking embodiment of Marx and Engels' view that the executive of the modern state exists to facilitate capital. Solomon worked as a political journalist at the CBC for a few decades and is now Canada's first Minister of Artificial Intelligence and Digital Innovation, leading Carney's rollout of the strategy. His job seems to be to pretend that he's representing the public in regulating AI while putting out this trust-me-bro corporate mission statement for oligarchs.
The day after the document dropped, Solomon gave a masterclass in word-salad salesmanship when he appeared on the CBC's Front Burner and was interviewed by Jayme Poisson about various elements of the strategy. In thirty minutes, he dodged nine questions, including: "Why does this strategy not put forward specific safeguards to prevent things like the Tumbler Ridge mass murder?" — a question referencing how ChatGPT influence was implicated in that tragedy. Also: "Why are no details offered around what that legislation could look like?" — to which Solomon responded that the proper place for details is in legislation, a genuinely circular answer. And: "Why no focus on AI job destruction? You're only talking about job creation." — to which Solomon pivoted to some Bank of Canada data showing no jobs had been lost yet.
Had there been more time, Poisson might have asked about how the US Cloud Act gives American authorities access to Canadian data regardless of server location. She might have asked how Gulf state funding of Canadian AI contradicts the strategy's democratic values framing. Solomon could have been asked about Palantir's Five Eyes contracts, which make all talk about data sovereignty hollow, because it's so deeply integrated with US security concerns. Solomon even leaked on air that Canada's Cyber Security Centre now has access to Claude Mythos — which means we're using a restricted US asset for national security while claiming sovereign AI infrastructure. Restricted, by the way, because Claude Mythos was understood by its developers to be potentially quite dangerous, and it was rolled back after an initial deployment.
There was also no question about whether Canada would support international regulation of AI in weapons systems and a binding treaty prohibiting autonomous lethal targeting — which is fairly germane, given that Solomon, as a liberal Zionist, refused to call Israel's military campaign a genocide, and that Canada has sold more than $30 million in military assets to Israel since that genocide started.
Solomon's favourite revolving-door keyword in both the interview and the document was "trust." When Poisson named a problem with AI, Solomon would essentially say: that's because trust must increase through accelerated adoption, so that more and more people can trust that the problems will go away. But he also used "trust" as a justification for deferring all specific safety mechanisms to forthcoming legislation — which, of course, won't be subject to real scrutiny or debate, because Carney has a majority and can slam things through.
Let me come back to the Marxist point. Solomon is on the show ostensibly to allay public concern — not because the public is concerned, mind you, but because public concern might obstruct the flow of capital. In Capital, Marx says that the state regulates the conditions of production — property rights, contract law, wage relations — to enforce the class system and guarantee that surplus value will be extracted from disciplined workers. In that sense, Solomon isn't corrupt. He's playing his role in what the modern state was built to do. The AI rollout is, as I said at the top of this episode, a perfect microcosm of the crisis of selling capitalism to the population it is immiserating.
Let me finish by circling back to the sovereignty theme, because there is actually a several-page section in the strategy document called "Supporting Indigenous Leadership in AI," which sounds good but which I think ties all of the deceptions together.
Here's a typical sentence from it: "Canada will work with Indigenous communities to ensure AI tools designed and built here protect Indigenous cultural and linguistic expression, support Indigenous self-determination over how AI is built and used in Indigenous contexts, and build domestic capacity to address the specific harms Indigenous Peoples face."
However, the Assembly of First Nations has already shown that the government has conducted no Nation-to-Nation consultation on AI, and that all previous AI legislation has infringed on First Nations rights in both process and substance.
The First Nations Information Governance Centre has a set of principles known as OCAP — Ownership, Control, Access, and Possession — and these are the established framework for Indigenous data sovereignty. Yet this concept does not appear in the new strategy document, although similar language does. Neither does the free, prior, and informed consent requirement articulated under the United Nations Declaration on the Rights of Indigenous Peoples, or UNDRIP. And that means Indigenous scholars are starting to speak out.
Dr. Jason Edward Lewis, Canada Research Chair in Computational Media and the Indigenous Future Imaginary at Concordia University, is the co-principal investigator of the Abundant Intelligences project — one of the largest Indigenous AI research initiatives in Canada, and a primary figure in Canadian Indigenous AI scholarship. He says: "Inclusion often means assimilation." You can see how that's working with the co-optation of a number of concepts into this document without reference to the frameworks that Indigenous people have already codified. Lewis says what we need is not assimilation, but transformation. He also says: "We don't have an AI ethics problem, we have an AI epistemology problem." Here he's alluding to the fact that generative AI is built on colonial business practices and so reproduces colonial hierarchies and risks. Along with many other Indigenous scholars, Lewis uses terms like "data colonization."
What they're saying is that the Carney model — which dictates who owns the infrastructure, who governs the data, who captures the returns — reproduces something very old and very familiar. But of course, what's true for Indigenous communities under colonial governance will be increasingly true for everyone: the workers whose labour data trains the models, the ratepayers whose electricity bills subsidize the buildout, the communities whose water tables are drawn down by cooling systems, and the artists whose creative work is harvested without compensation.
Lewis says that when intelligence and creativity are reduced to a rational, goal-seeking, self-serving process optimized for extraction, this is where we wind up. The Abundant Intelligences project advocates for data practices grounded in protocols of reciprocity and relationality: that data collection should serve the community it comes from, be obtained with ongoing consent rather than one-and-done extraction, and be governed by that community's own values.
That's all for today. Take care of each other, everyone.
