The Weekly Brief — Issue 10

Hello,

Canada is starting to set expectations for its data-centre build-out.

Ottawa has released national principles for how major data centres should be developed, including expectations around electricity costs, water and environmental impacts, local benefits, transparency and what Canada gets in return.

Plus: Quebec courts have drawn new lines around how judges should use generative AI, SOCAN is taking Suno to Federal Court, and AI is moving into governing policy in Quebec’s election campaign.

here is the week.

What happened

The federal government released Canada’s Responsible Data Centre Development Principles on September 3.

The framework sets five expectations for new data-centre development: projects should create lasting local benefits, avoid shifting project-driven electricity costs onto Canadians, minimize water use and environmental impacts, be transparent about local effects, and bring strategic value to Canada.

A group of major cloud, AI and data-centre companies have signed on to the principles, including Amazon Web Services, Anthropic, Bell, Cohere, Google, Meta, Microsoft, OpenAI and TELUS.

Ottawa says building more computing capacity in Canada can strengthen the country’s AI capabilities, digital resilience and sovereignty.

But the federal government is entering a debate that is already happening province by province.

New Brunswick opened consultation on its own proposed data-centre framework two days earlier. Alberta is working through how enormous proposed data-centre loads can connect to its electricity system. British Columbia and Saskatchewan have also been developing their own approaches to large AI and data-centre projects.

Why it matters

AI infrastructure is becoming a policy problem as much as a technology opportunity.

A large data centre can bring investment, construction and computing capacity. It can also require enormous amounts of electricity, new grid infrastructure, cooling and water, while affecting the communities that host it.

That creates questions governments increasingly have to answer before projects are built:

Who pays for new electricity infrastructure?

How much water and power will a project use?

What does the host community receive?

What information should companies have to disclose?

And what makes a project strategically useful to Canada rather than simply located here?

Ottawa’s principles now put those questions into a common national framework.

These principles are not a new national approval law for data centres. Electricity, land, water and project approvals still depend heavily on provincial, municipal, utility, regulatory and Indigenous processes. Ottawa is setting a baseline for responsible development, not replacing those decision-makers.

That distinction matters.

A national principle saying Canadians should not carry project-driven electricity costs is one thing. How that principle works when a utility assesses a real multi-hundred-megawatt project is another.

The next test will be whether the expectations announced in Ottawa meaningfully shape the projects now moving through provincial and local systems.

Your move

When you see the next big Canadian data-centre announcement, look for four things beyond the investment number:

Who supplies the electricity?

Who pays for the infrastructure needed to connect the project?

What does the host community receive?

And which parts of the announcement are actually binding?

That last question is particularly useful.

A law, regulation, regulatory decision, government framework, voluntary principle and corporate commitment can all sound authoritative in a headline.

They do not all carry the same weight.

Quebec courts set boundaries for judges using generative AI

Quebec’s Court of Appeal, Superior Court, Court of Québec and municipal courts have adopted common guidelines for judges using generative AI.

The guidelines make one boundary especially clear: judicial decision-making remains a human responsibility.

They say judges should not use generative AI to perform substantive legal reasoning, assess evidence or credibility, determine the outcome of a case, or produce the analytical reasoning behind a judgment.

Some secondary uses can be considered—for example descriptive, organizational or linguistic work—but only with human control and independent verification.

The guidelines also warn about hallucinations, bias, confidentiality risks and hidden instructions embedded in documents that could influence an AI system.

Why it matters: this is not a blanket ban on AI in Quebec courts. It is an attempt to define where assistance can end and accountable human judgment must begin.

That distinction will matter well beyond the courtroom as more professions decide which tasks can be supported by AI and which decisions should remain firmly human.

SOCAN takes Suno to Federal Court over AI-generated music

SOCAN filed a lawsuit against AI music company Suno on September 2.

The Canadian music-rights organization alleges that Suno’s platform generated and streamed outputs that replicate musical works in SOCAN’s repertoire without authorization or compensation.

Its Federal Court claim identifies a sample of 150 outputs.

The case has not been decided, and SOCAN’s infringement claims remain allegations.

Why it matters: Canada’s AI copyright debate is moving from theory into actual court cases.

The Canadian case adds another question to the AI copyright debate: what happens when an AI system produces and streams outputs that rights holders say cross the line?

Courts are increasingly being asked to define those boundaries.

AI moves into governing policy in Quebec’s election

Quebec’s election campaign is moving beyond rules for how political parties use AI during the campaign.

Parties are now starting to make commitments about how AI should be used and governed if they form government.

On September 2, the CAQ proposed accelerating the use of AI for repetitive tasks inside government and expanding its use in public administration.

The Quebec Liberals have separately called for an ethical framework for AI in public institutions and businesses, alongside training and workforce-transition measures.

The parties do not have identical approaches, and not every major party has published the same level of detail.

Why it matters: AI is becoming something voters may increasingly have to evaluate as ordinary public policy.

The questions are practical: where should government automate work, what protections should apply, how should workers be prepared, and who remains accountable when AI is used in public services?

That is a different conversation from whether a campaign used an AI-generated image.

It is about how a future government intends to govern with the technology.

When governments talk about protecting electricity ratepayers from data-centre costs, they are talking about who ultimately pays for the infrastructure needed to supply a very large new electricity customer.

A major data centre may need new transmission lines, substations, generation or other grid upgrades.

Someone has to pay for those things.

Depending on the rules, some costs can be charged directly to the project. Other electricity-system costs can potentially be spread more broadly across customers.

That is why the phrase “data centres must not shift electricity costs to Canadians” matters in Ottawa’s new principles.

It does not mean a data centre automatically raises everyone’s power bill.

It means governments and utilities have to decide how project-driven costs are allocated—and whether existing households and businesses should be protected from carrying costs created by a new development.

That is one of the most important details to watch as Canada’s AI infrastructure grows.

The next time you see an announcement about AI rules, take ten seconds and ask:

What is this actually?

A law?

A regulation?

A regulatory decision?

A government policy or framework?

A set of principles?

A voluntary code?

A company commitment?

Then look for the verb.

Must usually means something different from should, plans, expects or commits.

That simple check can tell you how much an announcement actually changes—and how much still depends on what happens next.

A calm, practical briefing on the AI developments that matter in Canada—delivered every Tuesday.

That’s the week. See you next Tuesday.

— Padge T.

AI Brief Canada is an editorially independent publication published by Groundshift Advisory Inc. and written by Padge T.

Groundshift Advisory helps Canadian organizations adopt AI responsibly. Sponsored, affiliated, or commercially connected content will always be clearly disclosed.

Issue disclosure: Nothing in this issue is sponsored, affiliated, paid, or commercially connected.

AI use: AI tools supported research, source discovery, outlining and drafting for this issue. All material was reviewed, edited and fact-checked by Padge T. before publication. Read our AI Use Disclosure.

Not affiliated with the Government of Canada.

Corrections or feedback: We welcome corrections and reader feedback. Email [email protected] or review our Editorial Policy.