TORONTO — The federal government’s long-awaited AI strategy fails to put guardrails around a new technology that’s making Canadians increasingly uneasy, critics of the plan said Thursday. 

The government released what it billed as a comprehensive strategy, titled “AI for All,” to develop the technology and infrastructure in Canada, boost adoption and protect people from the safety risk. But most of the safeguarding measures lacked detail or are slated to come out at a later date. 

“It’s really, in at least in its detail and where it specifies direction, an AI adoption strategy,” said Taylor Owen, the founding director of McGill University’s Centre for Media, Technology and Democracy, and a member of the government’s AI task force. 

The strategy promises new legislation to protect children online, including from AI, and to safeguard Canadians’ data privacy, for example. But it doesn’t say how it plans to do that, when the measures would take effect or how they would be enforced.

“That’s very different than the specificity that’s given for our compute rollout, which they can say to a fairly high degree of specificity what they’re going to do, to the dollar,” Owen said. 

There is far less emphasis on protections than on AI literacy, especially for children, said Helen Hayes, a fellow at Simon Fraser University that specializes in AI and digital rights. “Teaching young people how AI works doesn’t actually address whether those AI systems were designed responsibly in the first place,” she said. 

Carney said the government will table a bill in the next few weeks to address online safety for kids, but has not said publicly what will be included. The Logic reported last month that Ottawa plans to ban access to social media for kids, and is considering extending the policy to AI chatbots. Culture Minister Marc Miller has emphasized the importance of having a regulator that can hold companies accountable under new federal rules. 

The government hopes to table the bill before MPs break for the summer on June 19, Miller’s spokesperson Alisson Lévesque said in a statement.

The prime minister expects to discuss online child safety with his G7 counterparts at a summit in France later this month. Last week, AI Minister Evan Solomon signed an agreement with other G7 ministers that promised to take a unified approach. “Key priorities include implementing safety by design, using effective age assurance, enabling parental controls and improving education about online risks,” Innovation Department spokesperson Riyadh Nazerally said in a statement. 

The government hasn’t said when it expects to update its privacy laws for the AI era. 

Good rules will take time to develop, said Josh Tabish, senior director for the Chamber of Progress in Canada, a left-leaning tech lobby group. He said it’s important not to rush the process. “It’s great to see an AI strategy driven by pragmatism, not panic,” Tabish said in a statement.   

Recent polls suggest a large share of Canadians are uneasy about the technology and the threat it poses to jobs, privacy and the social fabric of the country. The news that the shooter that killed eight people in Tumbler Ridge, B.C., in February had exchanged concerning messages with OpenAI’s ChatGPT brought the issue to the political fore. 

In the time since the AI task force started working on the strategy in the fall, public sentiment about AI safety has changed, said Arvind Gupta, a University of Toronto computer science professor who sat on the panel. “I think it’s partially why the strategy took so long to develop,” he said. “They were thinking this all through.” 

The government doesn’t just need to win over the public, but to guard against AI that is used for nefarious purposes, Gupta said. The strategy includes a promise to work with emerging companies and allies to protect the country’s cyberdefences, but doesn’t include details.

CUPE, Canada’s largest union, accused the government of soft-pedalling protections against AI, and said it would like to see details about how the government plans to keep bots from replacing employees as the government pushes firms to adopt the technology.

“If the government doesn’t put up guardrails, workers will be managed by algorithms no one can see or challenge, and it will do real harm,” Mark Hancock, the union’s president, said in a statement. 

With files from Catherine McIntyre in Toronto