Alternative federal budget 2026-27: Artificial intelligence

This is from a larger publication, Alternative federal budget 2026-27: Bridge to independence

Introduction

So-called artificial intelligence (AI) software has taken the world by storm, driven by the proliferation of accessible chatbots and other tools that can perform tasks previously assumed to be impossible for machines. While AI tools show promise in some areas, they are already causing real harm in others. Canadians are justifiably skeptical of the pace and nature of the AI rollout.¹ In contrast to the federal government’s “all in” approach to AI development and deployment, the AFB takes a step back from the hype to lay out a more thoughtful and measured approach to the adoption of AI tools in Canada.²

Overview

The release of ChatGPT in fall 2022 ushered in an era of tech industry mania that has dwarfed even the delirium of the dot-com boom. Global investment in the artificial intelligence (AI) supply chain may exceed US$2.5 trillion this year.³ Technology companies are expected to spend half a trillion dollars on new data centres alone.⁴ These sums now account for a substantial share of the global economy and are projected to continue rising through at least the next decade.

At the core of this unprecedented private sector (and, increasingly, public sector) investment boom is the assumption that AI tools will deliver massive productivity gains across every part of the economy even though most commercial AI services are losing money today and most efforts to adopt AI at the institutional level have failed to produce returns.⁵ Unless these tools start delivering on their productivity promises, all this investment may amount to one of the largest financial bubbles in history with massive economic and social ramifications. Nevertheless, the narrative that AI represents the future of the economy remains sacrosanct. For many countries, including Canada, a fear of falling behind in the “AI race” is being used as a justification for large speculative investments and an absence of public interest regulation of the AI industry.

“Artificial intelligence” is, itself, a marketing term encompassing a variety of related technologies. These computer systems can make predictions or create new content based on sophisticated pattern matching, but they are neither artificial (AI systems are the product of human labour at every stage of development) nor are they intelligent (AI systems differ significantly from our own cognitive processes). Generative AI systems—which include the large language models (LLMs) such as ChatGPT, Gemini and Claude—are particularly prone to making basic errors because they are designed merely to predict words and do not have a persistent memory or an internalized sense of logic.

Nevertheless, modern AI tools are increasingly capable of producing outputs indistinguishable from many humans, which is already having profound consequences. Among other issues, unregulated AI use by students is undermining the public education system. AI-powered disinformation is running rampant on social media. False AI citations are polluting academic journals and the legal system. AI in the workplace is intensifying surveillance of workers and entrenching practices of algorithmic management. And while the large-scale automation of workers has not come to pass as quickly as many feared, there are early signs that the proliferation of AI tools is destroying jobs in specific sectors, especially junior level positions, which may be contributing to historic youth unemployment levels.⁶ Taken together, the proliferation of these tools represents an acceleration of the concentration of economic and political power among a handful of mostly-American tech companies at the expense of workers and the broader public.

The construction and operation of new data centres—the physical backbone of the AI boom—is creating a host of additional problems. A typical modern data centre uses as much electricity as 100,000 homes, which has one of two consequences. Either these facilities draw so much power from the grid that local power markets are distorted, which drives up rates for households, or they bring new electricity online, often in the form of gas-fired generators. The expansion of gas power associated with new data centres is already having a measurable impact on global greenhouse gas emissions.⁷

The data centre rush is also being used to force new infrastructure on vulnerable communities as well as ecologically-sensitive or agriculturally important land without adequate consultation or review. Once built, these facilities create vanishingly few jobs or local benefits.⁸ Many U.S. communities have begun to push back on data centres, and calls for a national moratorium are growing louder. Resistance to Canadian data centre proposals is also starting to foment.⁹ Canadian governments are not heeding the warning.

Indeed, since the abandonment of the controversial Artificial Intelligence and Data Act (AIDA) in 2025, the federal government has not introduced any AI legislation. In addition, Canada’s copyright and privacy laws are woefully outdated, leaving the AI industry essentially unregulated. Lower levels of government have started to step in, but public oversight remains piecemeal. The federal government recently advanced a new vision of “artificial intelligence for all” that acknowledges the need for safeguards, but five of the strategy’s six pillars are focused on AI investment and adoption.¹⁰ Instead of asking what problems AI can help solve, the government has taken the view that AI adoption is essential for its own sake. The AFB takes a different approach.

Actions

The AFB will accelerate the development and ensure the swift passage of a new, modernized Artificial Intelligence and Data Act that will give the federal government the power to introduce public interest regulation of algorithmic systems in both the public and private sector. To avoid the institutional capture that undermined previous legislative efforts, the new legislation will be administered by a body with a mandate limited to public interest regulation, independent of any department responsible for industry promotion or economic development (see Canadian AI and Data Commission below). The act will employ the precautionary principle to prohibit the introduction of AI tools into high-risk settings, such as education, health care or the delivery of public services, unless there is clear evidence that they will not cause harm. Where harms are already documented, the act will prohibit rather than pause. Prohibited applications will include predictive policing, facial recognition in public spaces, emotion recognition, human behaviour manipulation, and automated decision-making tools that affect the rights and livelihoods of the public, including in immigration and refugee proceedings. In addition, the act will include mechanisms to ensure publicly and commercially-available AI tools meet minimum standards for safety, reliability and transparency, including validation by independent third parties.

The AFB will allocate $250 million over five years toward the creation of the Canadian AI and Data Commission, an independent office of Parliament with a dual mandate of oversight and education. It will independently monitor the AI industry in Canada, empirically documenting both harms and benefits, and ensuring compliance with the new Artificial Intelligence and Data Act. It will do so without being constrained by the industry-adjacent innovation mandate that clouds existing government AI institutions. As part of its oversight function, the commission will serve as an ombudsperson for public complaints about AI harms. In its education function, the commission will develop a national AI literacy program in collaboration with trade unions and the public education sector as part of a broader participatory democratic process. This approach will serve as a deliberate counterweight to industry-led literacy initiatives that are increasingly being pushed on students, educators and institutions.

The AFB will ensure existing legislation is brought up to date for the AI era, including privacy, copyright and labour laws, to ensure users, creators and workers are protected from AI systems. Among other elements, updated legislation will ensure AI is not used for surveillance or data collection without the knowledge and, where applicable, the consent of workers or users. Building off the 2025 copyright consultation findings, updated copyright legislation will require consent from and compensation to creators whose work is used to train commercial AI systems. Updated privacy legislation will recognize Indigenous communities’ collective ownership and control over data about their lands, languages and cultures, requiring free, prior and informed consent for its collection and use. Updated labour legislation will prohibit algorithmic management practices for high-impact decisions such as hiring and discipline.

The AFB will require that all new data centres be powered with 100 per cent non-emitting electricity, and that at least 50 per cent of that power be provided through new renewable energy generation paid for by data centre developers. All data centres will be required to report to a public registry, which includes a detailed breakdown of energy use, water use, job creation and other local impacts.

The AFB will require the federal government to gradually divest itself of commercial AI services, especially those owned by U.S. companies that put Canadians’ sensitive data at risk. Canadian ownership and democratic control over the digital infrastructure and software underlying public services, including data storage, cloud computing and AI systems, requires that these tools be developed and managed in-house rather than contracted to foreign-owned vendors. The AFB will ensure the federal government has the technical capacity to manage the transition.

The Monitor

Our Schools/Our Selves

Alternative federal budget 2026-27: Artificial intelligence

Introduction

Overview

Actions

About the author

Alternative Federal Budget Working Group

Alternative federal budget 2026-27: Health care

Alternative federal budget 2026-27: First Nations

Alternative federal budget 2026-27: Health care

Alternative federal budget 2026-27: First Nations

Alternative federal budget 2026: Artificial intelligence

Budget fédéral alternatif 2026 : Intelligence artificielle

We’re fighting for change and your donation helps!

Contact Us

News & Resources

Regions

Projects

About Us

Contact Us

Ways to Give