OTTAWA — Amazon, Google, Meta and Microsoft are lining up against the Liberal government’s proposed artificial intelligence law, warning it would deter Canadian firms and consumers from more productive uses of the technology. 

Risky propositions: The Artificial Intelligence and Data Act (AIDA) would require organizations developing and deploying “high-impact” systems to identify and curb the risks of harmful or biased use. The government’s initial list of categories includes tools used in employment, health-care or judicial decisions.

But Ottawa should focus on the riskiness of specific AI applications rather than which industry or setting they’re used in, executives from the four tech giants told a parliamentary committee late Wednesday. “Not all risk is created equal,” said Amanda Craig, senior director of public policy at Microsoft, drawing a contrast between AI used for mortgage approvals and for more harmless applications like delivery routing. (Logistics isn’t currently one of the covered categories). 

Amazon, Google and Meta urged the federal government to assign AI rulemaking and enforcement to the regulators of specific sectors, rather than issuing rules for the technology in general: Health Canada for medical uses, the Office of the Superintendent of Financial Institutions for banking ones, and so on. The U.K. has taken that approach.

“Trying to do this as a horizontal piece of legislation is a little bit of a fool’s errand,” said Nicole Foster, director of global AI and Canadian public policy at Amazon Web Services, citing differences in the data that AI tools use and the people they affect. The innovation department has argued that a new law is required to fill in gaps in existing rules and help the public build trust in the technology.

On the block: In response to the Liberals’ other regulatory moves targeting tech, Google and Meta have withdrawn news links from their platform in the Canadian market, while Amazon threatened to pull its third-party marketplace. 

The compliance costs of AIDA would be “incredibly high,” per Rachel Curran, Meta’s head of public policy for Canada, who suggested the firm may not launch certain products in Canada as a consequence. She also cited its Fundamental AI Research team, led from Montreal by McGill University’s Joelle Pineau. “Over-regulation by the government is poised to drive that kind of activity out of the country,” Curran said. “I would hate to see us lose that.” Meta wants the government to drop content moderation and prioritization systems (read: the feed algorithms of social platforms) from the high-impact list.

Meanwhile, on Thursday Google finally launched Gemini, its rebranded generative tool, in Canada—a year after it went live almost everywhere else.