AI giant Anthropic is investing in Canadian AI research through partnerships with leading universities and the nation’s federally backed AI institutes, largely through credits for its Claude AI models.
Hugo Larochelle,
“Access to Anthropic’s models is a powerful accelerant for the work being carried out across Mila’s research community.”
Mila
Anthropic announced on Tuesday that it is committing $10 million CAD to fund research into “beneficial and responsible applications of AI.” The company said it’s partnering with Edmonton’s Alberta Machine Intelligence Institute, Montréal’s Mila, and Toronto’s Vector Institute, as well as CHEO, the Centre for Addiction and Mental Health (CAMH), Université Laval, University of Toronto (UofT), and the University of Saskatchewan—with more to be announced. Each institution will receive $1 million in Claude credits, and Anthropic won’t have control over research directions of findings, the company told BetaKit.
Across the universities, the company is providing credits for its widely used AI model, Claude, to support research and engineering. Amii, Mila, and Vector are also joining the Anthropic for Startups program, which could offer Canadian startups more Claude credits and resources.
“Access to Anthropic’s models is a powerful accelerant for the work being carried out across Mila’s research community,” Mila scientific director Hugo Larochelle said in a statement.
According to data that Anthropic released alongside the announcement, Canadians are overrepresented in their use of Claude AI models. While the country ranks eighth in the world for Claude usage (at 2.6 percent), it ranks second by AI usage index—an Anthropic measure for how much a country is using Claude relative to its working-age population. In that respect, Canada is just behind the United States, and above the United Kingdom and South Korea.
In a blog post announcing the funding, Anthropic noted that the University of Toronto and Université de Montréal both led research into neural networks—which underpin today’s generative AI systems—while researchers at the University of Alberta made headway in reinforcement learning. Canada’s three so-called “AI godfathers,” Geoffrey Hinton, Yoshua Bengio, and Richard Sutton, are affiliated with those universities, respectively.
Chris Olah, a Canadian researcher and Anthropic co-founder, nodded to the foundations of modern AI originating in Canada. “I was formed by that culture, and I’m proud Anthropic can support the next chapter,” he said in a statement. Olah studied at U of T for one year before dropping out, and later became one of the OpenAI employees who left to start Anthropic.
RELATED: Why the threat of Anthropic’s Mythos demonstrates the need for sovereign AI
The news comes as the Canadian government has said AI sovereignty is a pillar of its national AI strategy, which seeks to support domestic AI companies and help them scale up. Part of the sovereignty push has included providing startups with access to Canadian compute through initiatives like the AI Compute Access Fund. Canada only has one frontier model developer in Toronto’s Cohere, which could provide enterprise AI capabilities.
The feds were granted access to Anthropic’s Claude Mythos Preview model in early June through Project Glasswing, before the US instated a foreign-access ban on the model, then later lifted restrictions. In response to an inquiry about whether the Canadian Centre for Cyber Security’s model access had been restored, a Communications Security Establishment Canada spokesperson told BetaKit last week that Canada is still a participant in Project Glasswing, and that any further inquiries about the initiative could be directed to Anthropic.
Feature image courtesy Anthropic via Flickr. Image license under CC BY 2.0.