💰 $1.17 TRILLION Federal Debt  |  🏠 $817K Avg Canadian Home Price  |  📱 $54M ArriveCAN App  |  ⚖️ 2 Ethics Violations — First PM in History

The Daily Record

Accountability journalism the $600M government-subsidized media won't tell you.

Ottawa’s $150M LawZero AI Cheque Needs a Milestone Ledger

Safe AI can still be unsafe spending unless taxpayers see the receipts.

Editorial cartoon showing Ottawa handing LawZero a 150 million dollar AI cheque while taxpayers ask for milestones, compute contracts, IP safeguards and clawbacks.

Ottawa’s newest artificial-intelligence announcement is exactly the kind of file conservatives should scrutinize carefully: a fashionable cause, a large public cheque, and a thin public ledger.

On September 16, Innovation, Science and Economic Development Canada announced that Canada plans to invest CAD $150 million in Montréal-based not-for-profit LawZero through the Strategic Response Fund. Germany plans to put in EUR €100 million, subject to European Commission notification, and LawZero describes the combined commitment as up to CAD $300 million. The stated goal is “safe, sovereign AI” built around Yoshua Bengio’s “Scientist AI” approach.

There is nothing wrong with wanting safer artificial intelligence. If Canada is going to be in the AI race, it is reasonable to care about security, sovereignty, research talent and domestic computing capacity. LawZero may be doing serious work. Its model is presented as an alternative to autonomous AI agents, designed to reason transparently and provide evidence-based outputs rather than pursue goals of its own.

But a noble mission does not cancel the need for taxpayer discipline. The federal release says Canadian funding will support research and engineering talent and computing capacity. It says the project will create 360 full-time jobs in Canada and establish dedicated sovereign computing infrastructure with Hypertec and 5C. Those are measurable claims. They should not be left as press-release promises. If the jobs are real, count them; if the compute is sovereign, define it; if the technology is strategic, protect Canadian access in writing.

The basic problem is not the word “AI.” The problem is the blank space around the money. Canadians have not been shown the contribution agreement, payment schedule, milestone triggers, procurement terms for compute, job-verification method, Canadian IP rules, data-sovereignty safeguards, lobbying or conflict screens, or clawback language if promises are missed. When government chooses one organization for a nine-figure industrial-policy bet, the receipts should be public by default.

That matters even more because this is being wrapped in “sovereignty” language. Sovereignty is not a slogan. It means Canadians know where public money goes, who owns the resulting intellectual property, which contractors receive compute dollars, where data and model infrastructure sit, what Germany and the European Union can require, and what happens if the project changes direction.

The Carney government should publish a LawZero Milestone Ledger before the $150 million flows: every tranche, every deliverable, every job target, every compute contract, every Canadian ownership or access condition, every foreign-government condition, and every clawback trigger.

Safe-by-design AI may be a worthy research ambition. But safe-by-design spending requires something simpler: signed terms, public milestones and consequences when government promises do not become results.

The disclosure test: publish the LawZero contribution agreement, payment milestones, job-verification schedule, compute-procurement terms, Canadian IP and data-sovereignty safeguards, Germany/EU conditions, lobbying/conflict logs and clawback rules.
Sources

This article supports AI safety research in principle while arguing that nine-figure federal subsidies require public milestones, procurement transparency and clawback rules.