David Knight Legg: The U.S. AI panic is a gift to Canada - National Post

NP Comment David Knight Legg: The U.S. AI panic is a gift to Canada Join the conversation Share Close Share this article

Copy URL Facebook Linked In X (Twitter) Loading account options Sign In Manage Account Close account menu Sign Out An electrical distribution board inside the JN1 Equinix 4 megawatts capacity data centre in Germiston, near Johannesburg, on Oct. 1, 2026. (EMMANUEL CROSET / AFP via Getty Images) By David Knight Legg November 1, 2026 at 9:19 a.m. For several years, the most prominent public voices on artificial intelligence have been the CEOs of the largest American frontier labs: Sam Altman of OpenAI and Dario Amodei of Anthropic. Curiously, both have shifted away from AI optimism, instead delivering repeated warnings that advanced AI could produce catastrophic or even existential harm if development is not deliberately slowed or tightly supervised.

The result is a negative impression that AI is a looming threat rather than a productivity tool. In the United States and parts of Canada, this framing has fuelled opposition to new data centres even as Chinese competitors continue to scale rapidly.

Why did the two largest U.S. frontier labs suddenly turn negative on their own technology? Heading into IPOs, they appear to be seeking liability shields and regulations that would lock in a competitive advantage over smaller rivals.

But that campaign is not working on those who know the technology and the regulatory system best. Other labs and researchers treated the safety incidents as failures of design and testing, not as proof the technology itself required special protection.

The U.S. government agreed: as Treasury Secretary Scott Bessent put it in September, “it is humans who are responsible, not the AI.” He rejected any liability exemption that would let the large labs socialize risk while keeping the upside.

Canada has a similar regulatory choice to make in managing growing public anxiety about the AI doom narrative. It can listen to those demanding slow permitting and layered regulations that benefit the largest frontier-model incumbents. Or it can treat AI as an ordinary, fast-evolving technology. The latter makes more sense for three reasons.

First, competition is growing quickly, and regulation will need to be flexible. Open-source models are already demonstrating they can be downloaded, fine-tuned and run by universities, startups and public agencies at a fraction of the cost of the large closed frontier systems that want to define safety protocols or capability limits for everyone else. Onerous regulation will benefit the big incumbents and hurt smaller competitors.

Second, AI is a key part of strategic competition with China. Government rules that concentrate capability in a handful of protected U.S. firms aren’t healthy. China will ignore any such rules anyway. This will give its developers a practical advantage in speed and scale. The AI race will be won by innovations driven by the most vibrant and competitive market with the widest range of models. Letting large U.S. labs push for regulatory capture that slows western competitors is antithetical to the market forces the West needs to win.

Third, the existing large language models (LLMs) are only the beginning of AI technology. Large signal models and world models — systems that take continuous streams of visual, audio and sensor data rather than text — are now emerging as the next frontier of problem solving. They measure time, distance, speed and space, and use that data to control objects in the real world. They will make the current LLMs look like digital typewriters, and render their regulations obsolete.

Canadian policymakers must define the U.S. frontier labs and their AI doom narrative for what it is: an attempt at regulatory capture and structural competitive advantage that will set us back. They must reject it.

They must also create clear rules that secure public safety, have safeguards for dual-use and national security applications, and maintain clear liability rules for those who develop and deploy code. This will accommodate competition and the evolution of successive generations of models.

Our policymakers should also take advantage of the AI doom in the United States. Premiers can rapidly accelerate supercomputing infrastructure while U.S. governors declare years-long moratoriums.

Alberta has already made a strategic decision to do this. Within a week of New York Governor Hochul announcing a data infrastructure moratorium, Alberta announced a $13-billion, one-gigawatt Meta data centre project north of Edmonton.

That deal was the product of several years of deliberate policy work — mapping what hyperscalers need, establishing a bring-your-own-power model so developers fund their own generation and grid upgrades, and setting clear rules on water and industrial land use. These are ordinary industrial standards applied to ordinary (if large-scale) infrastructure. Saskatchewan Premier Scott Moe announced a $52 billion expansion of Bell’s AI infrastructure in September on similar terms.

Those policies will enable the data infrastructure that brings about all other economic development. Alberta now has more than 20 additional data centre projects before municipal councils, with dozens more in earlier stages of development, with UCP House Leader Joseph Schow indicating last week that government will enhance protections addressing power bills, grid stability and water use.

In short, the AI doom narrative of U.S. labs is a gift to Canadian premiers at a critical moment. Data centres are the digital equivalent of pipelines and power plants. They have the potential to run for decades, generating revenue and advancing manufacturing, health research, university labs and any business that depends on data. The computing models they support are no more inherently dangerous than advanced chemistry infrastructure.

Canada needs to act quickly and write effective and sensible rules that accommodate public safety and competition — and to build like crazy. Alberta and Saskatchewan have shown that there is a path to leading the supercomputing infrastructure boom.

Ignore the AI doom. Your grandkids will thank you.

Join the Conversation Postmedia is committed to maintaining a lively but civil forum for discussion. Please keep comments relevant and respectful. Comments may take up to an hour to appear on the site. You will receive an email if there is a reply to your comment, an update to a thread you follow or if a user you follow comments. Visit our Community Guidelines for more information.

You must be logged in to join the discussion.

The B.C. Conservatives must fix this province's culture