Should AIDA be included in Bill C-27?

Expert says move is not the best way to legislate artificial intelligence in Canada

Should AIDA be included in Bill C-27?

Incorporating the Artificial Intelligence and Data Act (AIDA) into the federal government’s proposed Bill C-27 is simply not the way to go around legislating AI, according to one expert.

Late in September, 45 civil society organizations, experts and academics released an open letter to François-Philippe Champagne, minister of innovation, science and industry (ISED), outlining key concerns about the proposition.

“The trouble with AIDA, as it's currently set up, is that it just hasn't gotten due consultation from a wide range of Canadian stakeholders. Minister Champagne has had a fair number of private consultations, primarily with AI industry people. But the relevant academics haven't really had a chance to weigh in. The public has had no chance to weigh in on the bill,” says Matt Hatfield, campaigns director at OpenMedia, in talking with Canadian HR Reporter.

“There's really been no scrutiny and stress testing of what they have. And it's a pretty problematic, inadequate bill currently as a result. So our strong preference would be that they remove AIDA from C-27 and propose it as separate legislation, and that they have a full public consultation on the bill before they pass it.”

Calls for employers to come up with policies and training on generative AI are getting louder amid the growing use of the emerging tech among employees across workplaces, according to a previous report.

Bill C-27— An Act to enact the Consumer Privacy Protection Act, the Personal Information and Data Protection Tribunal Act and the Artificial Intelligence and Data Act and to make consequential and related amendments to other Acts — is currently under committee consideration in the House of Commons.

Recommendation on improving Bill C-27

In the open letter, the stakeholders note: “It is clear that AIDA is not the AI bill people across Canada deserve. We need a reset, and if that is not to be, the bill requires substantial remediation that must incorporate the concerns we raise here.”

The signatories’ recommendations for government include the following: 

Recognize privacy as a fundamental human right: The stakeholders want Canadians’ privacy to have superiority over anything else when it comes to companies’ use of their data.

“We don't want commercial interests to either trump privacy interests, or to be considered equally important,” says Hatfield. “We think, ultimately, people's privacy rights are the most important thing. Of course, commercial interests are important, but they need to be bounded, reasonable, fairly narrow and adhere to respecting people's privacy rights.”

Remove AI regulation from ISED’s sole jurisdiction: Currently, under the bill, the AI commissioner who is providing most of the actual oversight of AI use is appointed by and reports directly to the minister of innovation, Hatfield explains. And the Ministry of Innovation is subsidizing “quite a bit of AI development”, and has virtual relationships with a number of AI developers.

“The idea that someone who is appointed by that person would be particularly tough in looking at what those companies are doing, essentially criticizing errors that they're making… I think that's an unreasonable expectation,” he says.

“We need to have that person not just reporting to ISED, engaging with ISED and engaging with other leading ministries, but really sufficiently independent to equally weigh all the considerations that go in there; not just sponsoring more innovation on AI use in Canada.”

Address poorly defined language in AIDA that create loopholes and a lack of enforceable rules: “There are special rules for high-impact AI systems under AIDA, but currently, in the text of the bill, it doesn't really define what those are,” says Hatfield.

“That's another area that Minister Champagne is now saying he is going to propose amendments around. But whether those definitions will be sufficient, that's to be determined. We don't actually have that text yet.”

Commit to far more active consultation with stakeholders beyond industry insiders: This, the stakeholders say, will ensure AIDA and subsequent AI rules are “well balanced and rights-protecting”.

However, with Bill C-27, there has been no consultation, says Hatfield.

“In the ordinary course of passing legislation on a complicated and relatively new topic, you would have a public consultation… The government has just skipped all of that. We've had no public consultation. Frankly, even some of the relevant stakeholders haven't had a great opportunity to hash out their ideas amongst themselves and test the different solutions that stakeholders have, because we haven't had that lengthy back and forth public process.”

Expand AI regulation to apply to both the public and private sectors, including government security agencies: Because AIDA has been put in Bill C-27, a private sector privacy bill, it, for the most part “does not apply to public bodies at all,” explains Hatfield. “No government agencies are regulated beta.”

This is a cause for concern, he says. “We would much prefer to see AIDA applied equally to public sector organizations as to private sector ones.”

Previously, one expert claimed that advancements around AI technology are pushing the world into “a period of huge uncertainty”. 

Latest stories