Canada has signed on to an international AI safety declaration committing companies and nations to greater information sharing and “early warning systems” to flag “serious cyber and AI control incidents.”

The France-led initiative drew initial endorsements from national governments — including Japan, Norway, the Netherlands and Bahrain — as well as major U.S.-based AI companies like Anthropic, OpenAI, Microsoft and Canada’s Cohere.

However, experts are skeptical that its voluntary nature and say without teeth, such agreements will not work.

Florian Martin-Bariteau, the director of the University of Ottawa’s Centre for Law, Technology and Society, said in an interview that while non-binding statements do send a signal about the Carney government’s approach to AI, they do not replace the need for serious regulation for the industry.

Martin-Bariteau said you can “go talk to the families in Tumbler Ridge” if you want to know how effective the AI industry’s voluntary standards are.

“It doesn’t work. The prime minister doesn’t seem to like actual regulators,” Martin-Bariteau said.

“I have nothing against anything in the statement, this is what we need … but if there is no teeth, if there are no people to do the follow-ups, the regulators, it will never work. Because it hasn’t worked before.”

The non-binding communiqué suggests agreement between governments and frontier AI companies in three general areas: a common risk management framework for the technology’s development, the principles of “responsibility and accountability” for all actors in the “AI ecosystem,” and increased information sharing to flag major threats.

On risk management, signatories said the would work to develop a public repository of non-sensitive information about AI risk assessments, with a standardized format to compare different models — something that would cut through the cryptic warnings AI companies often issue about the danger posed by their own models.

“Frontier models transcend national borders in their development, deployment and impact, and their opportunities and risks are therefore inherently global,” the statement, released by the Prime Minister’s Office on Monday, read.

“We all have a role in protecting our societies while enabling research, innovation and the wider use of AI.”

The declaration also stresses the need for governments to develop legal frameworks that give AI companies “certainty” about their development — as well as protect citizens “in case of large-scale damages.”

Prime Minister Mark Carney and his government have repeatedly touted the potential for AI to drive economic growth and innovation, launching the national “AI for All” strategy in June.

After well-publicized opposition to the construction of AI data centres in Canada and the United States, AI Minister Evan Solomon announced an agreement — again, a completely voluntary one — with leading AI companies to limit the environmental and noise pollution from new data centres, and to work with communities over concerns about the facilities’ water and electricity needs.

Ahead of last week’s federal budget, Carney announced that the federal government would make a $1.6 billion investment in Canadian AI company Cohere.

“We are not waiting for the future of AI to be decided somewhere else by those we don’t know or by whom we don’t trust. We’re building it right here in the right way,” Carney told reporters last month.

With a file from the Canadian Press.