The Canadian Legal Information Institute (CanLII) has filed a lawsuit against artificial intelligence (AI) startup Caseway AI, alleging Caseway violated its terms of use and infringed copyright by scraping 3.5 million records from CanLII’s database. The suit follows a failed cease-and-desist letter the institute sent to Caseway last month, demanding the startup delete and end use of any data allegedly owned by CanLII.
Caseway founder alleges suit is “an attempt to obtain a court order for data discovery without any concrete knowledge of infringement.”
In the lawsuit, filed in the British Columbia Supreme Court earlier this week, CanLII alleges that Caseway has created a business by wrongfully taking CanLII’s work through a bulk and systemic download from the database without permission or compensation. The suit adds that, in doing so, Caseway has breached CanLII’s terms of use and infringed on its copyright. The institute is seeking an injunction to restrain the startup from reproducing the allegedly stolen work, as well as monetary damages.
The AI startup, incorporated in both Dublin, Ireland and Vancouver, launched last month with a chatbot touted as a legal research assistant meant to fetch, explain, and summarize Canadian legal information and court decisions. Caseway claims its model avoids hallucinations by being trained to admit when it does not have access to an answer rather than trying to generate one.
CanLII, a non-profit organization founded by the Federation of Law Societies of Canada, provides an online database of court decisions, legislation, and legal commentary from all Canadian courts.
CanLII’s terms of use prohibit masking the origin of documents published on CanLII that are incorporated into another website, and forbids the bulk or systematic downloading of documents, including by programmatic means.
Caseway founder and CEO Alistair Vigier rebutted CanLII’s claims in a statement to BetaKit, saying that the court documents Caseway is trained on are public record and that neither he, nor anyone at Caseway, has “seen or accepted” CanLII’s terms of use. CanLII’s terms of use, hyperlinked on the bottom of every one of its webpages, states that every person who accesses the website consents to the terms of use as a user.
The lawsuit alleges that CanLII was alerted to copies of its content, amounting to over 120 gigabytes and 3.5 million records, being placed on a host that had the same IP address that was used to develop and host the Caseway platform. The filing adds that the investigation into how this occurred is ongoing. In response, Vigier alleged that the lawsuit is “an attempt to obtain a court order for data discovery without any concrete knowledge of infringement.”
“IP addresses are tied to individual computers, not to companies,” Vigier said, without directly addressing questions posed by BetaKit as to whether the company obtained any records or data using CanLII. “As CEO, I have never instructed anyone to copy CanLII’s content, nor has our company stored anything owned by CanLII,” he said.
In the filing, CanLII claims it has “reviewed, curated, catalogued, and enhanced” the publicly available court decisions, legislation, and secondary sources in its database at significant cost and effort. It claims its work includes structuring the case data, inserting and extracting keywords, summarizing court decisions, and providing original analysis.
Richard Gold, director of the Centre for Intellectual Property Policy and a law professor at McGIll University, told BetaKit that the copyright infringement component of the case would rely on whether Caseway made copies of the cases or simply looked at and analyzed them. Gold also said the test would be whether CanLII exercised “skill and judgment” in adding its enhancements to the public-domain (and therefore non-copyrightable) court documents it hosts.
“Correcting errors would not likely meet that [copyright] standard,” Gold said. “Adding hyperlinks, if done routinely by most people in the trade, would [also] not but if unique to CanLII may be. As for metadata, was it routine or did it require choice?”
Either way, when asked multiple times whether Caseway scraped any documents or data from CanLII, Vigier explicitly stated that Caseway “does not use any CanLII data or enhancements,” only the court decisions, but conceded that Caseway “still [doesn’t] understand what enhancements they suggest they created.”
Vigier added that he “never uses” CanLII because he prefers to use Caseway for his legal research, because it provides direct links to original court sources—including CanLII—so humans can verify Caseway responses.
The lawsuit is a Canadian twist on the torrent of legal action against AI giants like OpenAI, Microsoft, Anthropic, and Meta from intellectual property holders like newspapers, authors, artists, and some of the world’s largest record labels.
BetaKit has previously spoken with experts that assert academics, nonprofits, and early-stage AI startups are struggling to find material they can access for free to train their models, especially as a growing number of web publishers are attempting to bar AI web crawlers from scraping their content.
Colin Lachance, a legal AI consultant at PGYA and former president and chief executive officer of CanLII, echoed this sentiment in a LinkedIn post reacting to the lawsuit. Lachance said the lawsuit is reasonable, but notes the bigger picture is that Canada suffers for lack of bulk access to caselaw. In a Canadian Lawyer piece where he is quoted that, Lachance sympathized with Caseway, noting “There really is no other way for anybody to get bulk case law unless they have a relationship with the courts to receive it,” adding that to get access to Canada’s three million public court documents, the only tools available are CanLII and the library system of the Bar of Quebec.
Lachance said on LinkedIn that “Caseway clearly has zero understanding of what is permitted or possible in Canada when it comes to bulk access to case law.”
Vigier claimed that, since the lawsuit, Caseway has seen more than 200 lawyers and paralegals sign up for its monthly subscription service. He said that Caseway’s next major focus is on creating bespoke AI agents for specific legal workflows, such as generating drafts of a legal document after lawyers enter a few details about the people involved. Vigier added that Caseway doesn’t plan to use information uploaded by law firms to train its AI.
“People like Elon Musk often get sued because they are change agents,” Vigier said. “I see myself as a change agent in the legal industry, which is naturally litigious.”
Caseway has yet to file its defence and CanLII’s allegations have yet to be proven in court. BetaKit reached out to CanLII for comment but did not hear back by publication time.
Feature image courtesy Wesley Tingey via Unsplash.