The Alberta government under Premier Danielle Smith has turned to artificial intelligence as a tool in drafting new legislation. AI-generated drafting underpinned the Alberta Whisky Act, a piece of legislation laying down fresh labelling requirements intended to raise the profile of distilled spirits produced in the province.
Proponents envision the product earning a regional distinction comparable to the protected status Kentucky bourbon holds within the spirits world. No Canadian province or territory had previously turned to artificial intelligence when drafting legislation, making Alberta's approach a reported first of its kind within the country.
Two reports on the matter, released by Fasken Martineau DuMoulin LLP within a span of weeks, have drawn wider attention to both the potential advantages and the attendant risks that governments face when turning to artificial intelligence as a tool in drafting legislation. The opening entry in Fasken's "Distilled Law" series established that Alberta's legislative experiment drew notice from audiences well beyond food and beverage law — policy analysts, technology advocates, and intellectual-property practitioners among them — all watching to see how government adoption of AI might translate into practice, a thread the series picked up on Sept. 15, when its second instalment turned to the question of what role AI should play in the drafting of legislation.
Dale Nally, serving as Minister of Service Alberta and Red Tape Reduction, noted that AI tools played a role in the early drafting stages of the legislation — a development he described as a positive step forward for the province — while making clear that the government had no intention of ceding authorship to the technology or stepping back from its responsibilities in the process. Nally told journalists earlier that year that every stage of the process would be subject to scrutiny, and that artificial intelligence had been applied to the data to extract the foundational elements from which the work could proceed.
He noted that while a future in which artificial intelligence drafts legislation entirely on its own was not one he foresaw, the technology was nonetheless expected to serve as one instrument among several that lawmakers draw upon when crafting laws.
A second report from Fasken identifies constraints on the use of AI tools in drafting legislation, observing that while such technology can handle tasks like consolidating large volumes of information and identifying factual inaccuracies, the process of writing laws also calls for an understanding of government policy goals and the practical effects on people's lives. The report further notes that because AI systems are shaped by their training data, they carry the potential to embed assumptions or preferences misaligned with the specific circumstances of a given legislative initiative or weighted toward approaches already prevalent in that data, making continued human involvement in the oversight process a matter of practical necessity rather than preference, according to the report's findings.
The Fasken report states that legislative counsel and policy-makers retain the ability to examine and evaluate every dimension of how AI tools are applied to the drafting of legislation. That finding sits alongside Alberta's separate deployment of an AI-enabled wildfire forecasting platform, which draws on historical fire records, weather patterns, and ecological data, as documented in a 2025 report published by the Organisation for Economic Co-operation and Development.
According to the same OECD research, AI adoption by governments has been concentrated in streamlining internal administrative functions and delivering services to the public, with lawmakers drawing on the technology for policy development far less frequently. On the matter of drafting legal texts, the OECD's report notes that AI systems are capable of producing initial versions of documents by drawing on existing legislation and template structures, a process the organisation says promotes adherence to established standards while reducing the time and resources involved, with the added capacity to cross-reference new drafts against current laws so that potential conflicts and human errors can be identified before finalisation.
