Alberta used AI to draft its whisky act, giving Canada’s legal community a buzz

Alberta Premier Danielle Smith’s government has taken a spirited approach to using artificial intelligence to craft new legislation.

It used a dram of AI to draft its landmark Alberta Whisky Act, which sets out new labelling standards to boost the brand power of the Western province’s distilled drink. The idea is to give it a regional identity similar in stature to Kentucky bourbon.

But that marketing strategy isn’t the most potent part of this story. Alberta’s pioneering use of AI is believed to mark the first time that any Canadian jurisdiction has used the technology for lawmaking purposes. So it is easy to see why the country’s legal community thinks that’s neat.

Law firm Fasken Martineau DuMoulin LLP has published two reports on this issue in recent weeks, sparking a broader conversation about the opportunities and risks for governments that use AI when creating legislation.

The first instalment of Fasken’s “Distilled Law” series argues that Alberta’s trailblazing move has garnered attention far beyond the food and beverage industry, piquing the interest of policy makers, technophiles, intellectual-property lawyers and other legal experts curious about government use cases for AI.

“The Whisky Act highlights a broader issue: what role should AI play in the development of legislation?” states Fasken’s second segment in the series, published on Sept. 15.

One-fifth of MPs’ speeches in the House of Commons are written with AI help, analysis finds

Dale Nally, Minister of Service Alberta and Red Tape Reduction, has said the government used AI in the initial drafting of the legislation, calling it “a first good step” for the province.

“We said all along that we’re not going to let AI write the legislation and we’re just going to take a seat at the back of the room. We said that there’s going to be eyes on this every step of the way,” Mr. Nally told journalists earlier this year.

“So we used AI to analyze the data and to really come up with the building blocks from which we can work from. And so we’re pleased with that.”

Alberta’s government is certainly breaking new ground with its novel use of AI. Even so, Mr. Nally appears to be clear-eyed about the technology’s potential pitfalls.

“We’re not really anticipating a world where AI writes all of our legislation alone,” he said. “But we certainly anticipate one where AI can be one of the tools” used to write legislation.

That is the most significant insight that one can glean from Alberta’s foray into AI-assisted lawmaking. Actual human beings never relinquished their responsibility for oversight.

Fasken’s analysis appropriately highlights the “limitations of AI-assisted legislative drafting,” even if the technology can tackle tedious, time-consuming tasks such as synthesizing large amounts of information and flagging factual errors.

It stresses that drafting legislation also “requires contextual knowledge and judgment” about a government’s policy objectives and any real-world consequences for citizens.

“There is also a risk of bias: because AI systems reflect the data on which they are trained, they may import assumptions or priorities that do not fit the specific conditions of the legislative initiative or that favour particular approaches reflected in such data,” Fasken’s second report states.

“For these reasons, human oversight remains essential. Legislative counsel and policy-makers must be able to review and assess all aspects of AI-assisted legislative drafting.”

Who is legally responsible when AI agents go rogue?

In other words, government officials and bureaucrats cannot behave like automatons while using AI to draft laws. Human discernment provides a required dose of sober second thought.

Still, Alberta’s experimentation is promising because it points to more practical AI applications for governments, which are known for their plodding productivity.

For its part, Alberta also uses an AI-enabled wildfire prediction system that draws on historical fire, weather and ecological data, according to a 2025 report by the Organization for Economic Co-operation and Development.

That same OECD study found that governments mostly use AI to improve their “internal operations” and for “public service delivery.” Legislators rarely use it for policy-making purposes.

With respect to drafting legal texts, the OECD notes that “AI systems can generate preliminary drafts” by using templates and existing laws.

“This process helps ensure adherence to standards while saving time and resources,” its report states. “Additionally, AI can cross-reference new drafts with current laws, identifying potential conflicts and minimising human errors.”

Governments, therefore, face the challenge of mitigating the technology’s risks without missing out on its benefits.

Alberta’s use of AI to draft its whisky law is proof of the tech’s evolving potential. But governments must use it with a peg of prudence or risk finding themselves on the rocks.