Skip to content
AI

Canada and Germany Put Up to CAD 300 Million Behind LawZero

Ottawa commits CAD 150 million and Berlin up to €100 million to Yoshua Bengio's non-profit, which will open a Berlin office and build dedicated compute in Canada for its Scientist AI project.

By
· Updated 3 min read
inLinkedIn𝕏Post
Palais des congres de Montreal 33
Palais des congres de Montreal 33 · Jeangagnon · CC BY-SA 4.0 · via Wikimedia Commons

Canada and Germany announced funding of up to CAD 300 million for LawZero, the Montréal non-profit founded by Turing Award winner Yoshua Bengio, at the ALL IN conference in Montréal on 16 September 2026. The Canadian government is contributing CAD 150 million, according to CBC News, which reported that this is the first federal money LawZero has received. Germany's digital minister, Karsten Wildberger, announced plans to provide up to €100 million, subject to approval by the European Commission, The Canadian Press reported.

LawZero's announcement says the money will expand its international research team, pay for computing infrastructure and open an office in Berlin. The compute will be a dedicated, sovereign installation in Canada built with two Canadian companies, Hypertec and 5C.

A model designed to have no goals

LawZero's flagship project, Scientist AI, starts from a premise that runs against the direction of the large labs. Frontier systems are increasingly built to act autonomously and pursue goals. LawZero says Scientist AI is being developed to reason transparently and give evidence-based answers without goals of its own. The organisation expects it to act as a guardrail on today's frontier models, speed up scientific work and improve understanding of AI risk.

The safe-by-design model offers an alternative to potentially uncontrollable autonomous AI agents.

Karsten Wildberger, Federal Minister for Digitalization and Government Modernization, Germany

The announcement refers to recent incidents of systems crossing safety boundaries, and there has been a run of them. Reuters reported on 9 September that Anthropic had disclosed a fourth case of a model hacking external systems during testing. Prime Minister Mark Carney called this week, in an interview with Bloomberg, for a global, non-industry body to oversee AI risks, CBC reported. Bengio told the conference he was acting "with the urgency that I feel in my bones". Evan Solomon, Canada's AI minister, rejected the idea that the grant was a reaction to the news: "We are here today not because we are responding to yesterday's headline, but because we are responding to the philosophic position that we have taken, which is to build responsibly and safely."

Safety money with a sovereignty label

Both governments presented the grant as industrial policy as much as safety policy. Solomon's statement in the release speaks of strengthening Canada's sovereign AI capacity so that the next generation of AI is built "on Canadian terms". Wildberger's refers to technological sovereignty for both countries. The announcement came hours after Canada's Cohere and Germany's Aleph Alpha signed their merger agreement, a deal CBC quoted Vass Bednar of the Canadian Shield Institute describing as "arguably capacity building".

LawZero already had philanthropic backing from the Gates Foundation, Coefficient Giving, Schmidt Sciences, the Future of Life Institute, NVIDIA and Jaan Tallinn, according to its release. Its board is chaired by Maria Eitel, and its advisory council includes the former prime ministers Jacinda Ardern and Stefan Löfven.

Small sums by frontier standards

The amounts need perspective. Germany's share of up to €100 million is a thirtieth of the €3 billion that Mistral raised last week, Reuters reported on 16 September. The honest reading is that the grant buys LawZero time, researchers and hardware, not a place at the frontier. Its value will depend on whether Scientist AI works as a guardrail that other developers, or their customers, choose to run, which makes it a product question as much as a research one.

The most direct scepticism came from Jaxson Khan, an AI policy expert at the University of Toronto's Munk School of Global Affairs, who told CBC: "There's no silver bullet to this problem right now." He still called the investment worthwhile. LawZero's announcement gives no timeline or benchmark results for Scientist AI, and Germany's contribution is not final until the European Commission approves it.

What happens next?

  • The European Commission must approve Germany's contribution of up to €100 million before it is final.
  • LawZero plans to open a Berlin office and build dedicated compute in Canada with Hypertec and 5C.
  • Watch for the first published results or deployments of Scientist AI as a guardrail on other models.

Sources & references

  1. 01LawZero receives a commitment of up to $300M in joint funding from Canada and GermanyLawZero via PR Newswirecompany16 September 2026; includes ministers' statements
  2. 02Canada is investing $150M to make AI safer. Will it work?CBC Newsnews16 September 2026; Canadian share, Solomon, Bengio, Khan and Bednar quotes
  3. 03Montreal AI research institute denounces violence after office vandalizedThe Canadian Press (via Lethbridge News Now)news16 September 2026; German share subject to European Commission approval
  4. 04Anthropic discloses fourth AI hacking incidentReutersnews9 September 2026
  5. 05Cohere, Aleph Alpha combine to target enterprise AI marketReuters (via Yahoo Finance)news16 September 2026; Mistral €3 billion raise
Published 17 September 2026 · Updated 17 September 2026 · Report a correction · How we use AI
inLinkedIn𝕏Post

More from Artificial Intelligence

View all
AI

California Builds the Plumbing for AI Audits

Governor Newsom signed SB 813 and AB 1405 on 9 September 2026, establishing a framework for independent verification organisations and a state registry of AI auditors with independence rules, backed by both Anthropic and OpenAI.

5 min read
AI

GPT-6 Astra Crosses the Line OpenAI Drew for Itself

OpenAI released GPT-6 Astra on 3 September 2026, claiming saturated frontier benchmarks and state-of-the-art computer use. Its system card marks it Critical for cybersecurity under the Preparedness Framework, so exploit-class capability goes only to vetted defenders while general access rolls out broadly.

6 min read
AI

Humanoids Clock In: What the First Factory Contracts Reveal

Global humanoid shipments jumped to between 13,000 and 18,000 units in 2025, led by Chinese vendors, but most went to entertainment, education and data collection. The first binding factory contracts in Europe show where commercial deployment actually starts: box handling, battery assembly and robot-as-a-service pricing.

7 min read
AI

Open Weights Just Got Bought

In one summer, the largest open-weight models ever released arrived from Moonshot, DeepSeek and Alibaba, and the two platforms through which developers find and route them were acquired by NVIDIA and Stripe. Open weights are strategic infrastructure, and the strategy now has owners.

7 min read