OpenAI cannot dodge its Canadian copyright case after all. The Court of Appeal for Ontario dismissed OpenAI's appeal on November 13, ruling that Canada's largest news publishers may keep suing OpenAI in Ontario, accusing it of using their journalism to train ChatGPT without permission. It is the first Canadian case dealing with generative AI training and copyright, and this time the court decided only "where the case is fought" — it has not yet touched the core question of whether OpenAI actually infringed anything.
What the Court Actually Decided
OpenAI's strategy was a jurisdictional challenge: as a Delaware company, it argued that any infringement, if it happened, took place outside the United States, that Canadian courts had no authority over it, and that the case belonged in an American courtroom. The Ontario Superior Court of Justice rejected that argument on November 7, 2025, holding that Ontario does have jurisdiction. OpenAI appealed, and on November 13 a unanimous three-judge panel led by Justice B.W. Miller dismissed the appeal, upholding the lower court's ruling and ordering OpenAI to pay $100,000 in legal costs to the publishers. The court also noted that OpenAI had filed no evidence showing that a Canadian judgment would conflict with existing American law and complicate enforcement.
| Date | Development |
|---|---|
| 2024 | Six publishers sue OpenAI in the Ontario Superior Court of Justice |
| November 7, 2025 | First-instance ruling: Ontario has jurisdiction; OpenAI's objection dismissed |
| November 13, 2026 | Court of Appeal dismisses OpenAI's appeal; $100,000 in costs ordered |
| Next | Case returns to the Superior Court for a hearing on the merits; no date set |
Who Is Suing, and What They Allege
The plaintiffs are half of Canada's news industry: Toronto Star Newspapers, Metroland Media, Postmedia Network, The Globe and Mail, The Canadian Press, and CBC/Radio-Canada. The six organizations sued together in 2024, alleging that OpenAI copied large amounts of their journalism without permission or payment to train the large models behind ChatGPT and profited from it — copyright infringement, breach of website terms of use, and unjust enrichment. The publishers have also filed a separate, similar lawsuit against Anthropic. In the United States, a parallel fight is under way: the Ross case, in which a request for a full-court rehearing challenges a ruling on AI training copyright, is covered in our report.
What This Ruling Means
For the publishers, the win is procedural: the case does not have to move to the United States, and it can be fought in their own courts. For OpenAI and the wider AI industry, the signal is worth noting — a foreign AI company cannot escape another country's courts simply because its "servers and company are in the United States"; the Canadian court treated "the business and the alleged wrongdoing having a real connection to Ontario" as the basis for jurisdiction. But the outcome on the merits is far from decided: whether OpenAI actually infringed, whether training counts as fair dealing, and how damages would be calculated all await the merits hearing. The publishers have won a ticket to keep litigating, not a verdict.
