
Key Highlights:
- The Ontario Court of Appeal has denied OpenAI’s appeal, allowing five Canadian news publishers’ copyright lawsuit against the company to proceed in Ontario.
- The publishers allege OpenAI used their journalism to train ChatGPT and commercially benefited from their work without permission or compensation.
- Unlike the US fair use framework, Canada applies fair dealing provisions and has no specific text-and-data-mining exception for AI training in its Copyright Act.
OpenAI faces a major legal drawback in Canada over allegations that it used copyrighted news articles without authorization to train its AI models. The dispute, brought by five leading Canadian news organizations in November 2024 has reached another milestone after the Ontario Court of Appeal denied OpenAI’s plea to have the case heard in the US. The ruling keeps the case in the Canadian administration and could make it a key test of how the nation’s framework applies to generative AI. However, the court has not yet decided whether OpenAI actually infringed copyright.
Why Are Canadian Publishers Taking OpenAI to Court?
The lawsuit was filed in November 2024 by five major Canadian news firms: The Canadian Press, Torstar, The Globe and Mail, Postmedia and CBC/Radio-Canada. The publishers accused OpenAI of scraping large amounts of journalistic content from their websites and using the material to develop ChatGPT and its other products without obtaining permission or providing compensation. obtaining permission or providing compensation.
The publishers’ 84-page statement of claim alleges that OpenAI accessed and copied articles from their websites, web-based applications and third-party partner websites. They also claim that the company used copyrighted material in its GPT models and continued accessing and reproducing their content in response to user prompts. Beyond copyright infringement, the legal filing raises allegations of circumventing technological protection measures, including those used to restrict unauthorized access, and breaching the publishers’ website terms of use.
The publishers are seeking financial damages and a permanent injunction that would prevent OpenAI from using their copyrighted material without consent. They argue that journalism requires substantial investment in reporting, editing and publishing, and that AI companies should not be able to commercially benefit from this work without compensating its creators.
What Does the Canadian Court Ruling Mean for AI Copyright Rules?
OpenAI had argued that the lawsuit should be heard in the United States because the company is headquartered there and the alleged web crawling and AI training activities took place outside Ontario. The Ontario Superior Court rejected that jurisdictional challenge in November 2025, allowing the claims to proceed in Canada. OpenAI subsequently appealed that decision, but the Ontario Court of Appeal dismissed its appeal on November 13, 2026.
In its decision in Toronto Star Newspapers Limited v. OpenAI, Inc., 2026 ONCA 814, the appeal court found no reason to overturn the lower court’s conclusion that Ontario had a real and substantial connection to the dispute. The judges also rejected OpenAI’s argument that the United States was a clearly more appropriate forum. The court noted that OpenAI had not established that a Canadian judgment would conflict with existing US laws.
The distinction between Canadian and US copyright law is central to the case. The United States has a fair use doctrine that allows certain uses of copyrighted material without permission, depending on factors such as the purpose of the use and its effect on the market for the original work. Canada instead uses a fair dealing framework, which recognizes specific purposes, including research, private study, education, parody and satire.
Canada’s Copyright Act does not currently contain a dedicated exception that expressly permits AI companies to use copyrighted material for text and data mining or model training. However, this does not automatically mean that every use of copyrighted material for AI training is unlawful. OpenAI may still seek to rely on existing legal exceptions, including fair dealing, and the court will need to examine the specific facts and arguments before reaching a decision.
The appeal ruling therefore settles a procedural question rather than the central copyright dispute. The next stage will examine whether OpenAI’s alleged collection, reproduction and use of Canadian publishers’ content violated Canadian law and whether any available legal defences apply. The outcome could provide important guidance for AI developers, publishers and other copyright holders operating in Canada.








