Why the EU and Canada are betting on an AI ‘middle powers’ play

Por Pieter Haeck — China – POLITICO

BRUSSELS — European Commission President Ursula von der Leyen hailed a new “tech alliance” between the EU and Canada in her State of the Union. During his own speech Thursday to the European Parliament, Canadian Prime Minister Mark Carney declared that the two sides should “join forces” on AI safety and pooling compute power.

Yet even before the public displays of connection in Strasbourg, there’s already plenty bringing Canada and Europe together on the AI front. Both lag behind the U.S. and China in AI development and compute power, and thus lack the political clout to set the direction and pace of global regulation. Together, however, Brussels and Ottawa could maybe — just maybe — compete for leadership in the development and policing of this emerging and powerful technology.

It is Carney’s “middle powers” pitch in action.

In an address to the World Economic Forum earlier this year, the Canadian prime minister argued that such second-tier powers as the EU and Canada “must act together” to counter U.S. and Chinese power. Von der Leyen on Wednesday backed the idea, describing how Canada and the EU should team up to build tests and monitoring regimes to keep powerful AI models in check.

Already, the Canada connection has helped shape Europe’s AI landscape. Canadian computer scientist and AI guru Yoshua Bengio was a key architect of the EU AI rules, having drafted a set of security practices for AI frontrunners in 2024. The European Commission and Canada then signed an AI-focused collaboration which included measures to boost AI adoption late last year.

Perhaps the most concrete sign of the pairing was the April announcement that Canadian AI player Cohere would merge with German counterpart Aleph Alpha in a multi-billion-dollar team-up.

This EU-Canadian multinational, with offices in Toronto, New York, London, San Francisco, Montreal, Paris, Seoul and Berlin, is positioned as more than just a business deal — particularly amid the repeated reports of rogue agents and Washington and Beijing falling into a kind of AI arms race.

“Europe and Canada understand that there needs to be a second democratic rail in the race for AI,” Aidan Gomez, CEO of Cohere, said in an interview.

He argued that the current “bifurcation” of the AI world into one part led by a democracy, the U.S., and one led by an autocracy, China, is a “dangerous one for the future.”

Offering an alternative was “the principal motivation” for the team-up with German AI research lab Aleph Alpha, Gomez said. The joint company is reportedly valued at $20 billion, and the transaction should close soon.

Gomez has a British mom and a Spanish dad, but grew up in Canada, embodying the EU-Canadian link like few others.

This summer’s barrage of incidents involving agents from OpenAI and Anthropic escaping their test environments and hacking into other companies highlights why the EU and Canada need to develop advanced capabilities of their own.

“The question is whether Europe, the EU, Canada will have and be able to develop these systems themselves to defend their critical industry, their markets, their sensitive workloads,” Gomez said.

Matter of survival

“China and the U.S. are,” he argued. “So we need to invest to create a similar capability because it will be fundamental to competitiveness and relevance on the global stage.”

After the recent call by top American CEOs to collectively slow down the pace of AI development, Gomez published an essay Sunday accusing Silicon Valley of behavior that is “a cartel by any other name.”

Bengio, who some consider one of the “godfathers” of AI, seconded Gomez’s push, calling technological sovereignty “a matter of economic and political survival.”

“As AI becomes a key driver of competitiveness, Canada-EU cooperation on secure, reliable AI systems will be essential to safeguarding our shared sovereignty,” he said in a statement.

Bengio is a voice the EU pays attention to. He was one of the academic experts tasked by the European Commission to write a set of best practices on safety and security practices that operationalized a part of the EU’s flagship AI law.

At the end of last year, the European Commission and the Canadian government laid the groundwork for deeper collaboration on AI by signing a memorandum of understanding. That document committed both sides to “accelerate the adoption of AI and foster its responsible development and use.”

Collaboration topics include AI standards, AI skills development, scientific collaboration, and the adoption of AI in strategic economic sectors.

This article has been updated.

Fonte: China – POLITICO