Canadian Court of Appeal Rejects OpenAI Motion to Dismiss News Publisher Copyright Class Action
The Ontario Court of Appeal clears the path for Canadian newspapers to proceed with litigation alleging unauthorised scraping of copyrighted journalistic content for ChatGPT training.

By Rachel Lim, Venture, Markets & Policy
, updated 6 min read
The Court of Appeal for Ontario has dismissed an appeal filed by OpenAI, allowing a sweeping copyright infringement class action brought by a consortium of major Canadian news publishers to proceed to trial.
The litigation, initiated by media organisations including The Globe and Mail, Torstar Corporation, Postmedia Network, The Canadian Press, and CBC/Radio-Canada, alleges that OpenAI systematically scraped and reproduced millions of copyrighted news articles without authorization or compensation to train its commercial artificial intelligence models.
The appellate court's unanimous decision upholds a lower court ruling that dismissed OpenAI's jurisdictional challenge and request for summary dismissal, establishing that Canadian courts possess subject-matter jurisdiction to hear claims involving foreign technology companies whose automated web crawlers harvest Canadian digital content.
The Territorial and Copyright Claims
OpenAI had argued that Canadian courts lacked territorial jurisdiction over the dispute because its data collection servers, computational training clusters, and corporate headquarters are located in the United States.
The three-judge appellate panel rejected this argument, ruling that the alleged copyright infringement occurred within Canadian borders when automated crawlers accessed Canadian web servers, circumvented technical access controls (such as paywalls and robots.txt directives), and reproduced digital works stored in Canada.
Under the Canadian Copyright Act, unauthorized reproduction of a substantial part of a copyrighted work constitutes direct infringement. Unlike United States copyright jurisprudence, which relies on a broad and flexible fair use doctrine, Canadian copyright law provides a narrower, closed-list framework known as fair dealing.
Fair dealing requires that any unauthorized use fall strictly within enumerated statutory categories, such as research, private study, education, parody, or news reporting, and satisfy a stringent six-factor fairness test. Legal analysts indicate that commercial model training by a venture-backed technology corporation faces significant hurdles qualifying under Canadian fair dealing exemptions.
Circumvention of Digital Paywalls and Terms of Service
The publishers' statement of claim presents technical evidence alleging that OpenAI's automated scraping infrastructure systematically bypassed subscription paywalls and anti-scraping mechanisms implemented on Canadian news websites.
The plaintiffs assert that ChatGPT frequently reproduces verbatim paragraphs, unique reporting disclosures, and proprietary investigative journalism when queried by Canadian subscribers. The publishers argue that by reproducing reporting verbatim while simultaneously competing for digital readership and advertising revenues, the artificial intelligence developer directly cannibalizes the commercial value of their underlying editorial assets.
The lawsuit seeks statutory damages under Section 38.1 of the Copyright Act, which allows copyright holders to claim up to $20,000 per infringed work, alongside an injunction prohibiting OpenAI from using Canadian news content in future model training runs without a commercial licensing agreement.
Broader International Licensing Precedents
The Ontario ruling arrives as international publishers negotiate bilateral commercial licensing agreements with frontier AI developers.
OpenAI has previously signed content partnership and licensing agreements with global publishing houses, including News Corp, Axel Springer, Le Monde, and Prisa Media. However, Canadian publishers have taken a unified adversarial stance through class litigation, seeking to establish a binding legal precedent that forces artificial intelligence companies into collective licensing agreements similar to Canada's Online News Act framework.
Publishers argue that collective legal action is essential to ensure that regional media organisations receive equitable compensation, rather than allowing tech platforms to strike fragmented deals with select global conglomerates while leaving local journalism underfunded.
Next Procedural Steps and Industry Impact
With the appellate court dismissing OpenAI's preliminary objections, the litigation moves into the formal class certification and documentary discovery phases.
During discovery, the plaintiffs will seek access to OpenAI's internal training datasets, data filtering logs, and crawler access records to determine the exact volume of Canadian journalistic content ingested into GPT-4, GPT-5, and subsequent reasoning architectures.
A spokesperson for OpenAI expressed disappointment with the appellate ruling, maintaining that the company builds its models using publicly available data in accordance with international legal principles, and indicated that the firm will continue to defend its development practices in court.
Sources
- 1Court of Appeal for Ontario Judgmentsontariocourts.ca