AI Momentum 100.8 -1.7%GitHub Pulse 4636 -14.0%Capital Flow $207.4B +4676.9%Hiring Heatmap 0.0 new
Advertisement
Regulation & Policy

OpenAI loses its bid to move Canadian news publishers' copyright suit to the US

OpenAI has lost its attempt to move a landmark Canadian copyright case over AI to the United States. The Canadian Press calls it the first Canadian case to address training AI on copyrighted content. On Friday, the Court of Appeal for Ontario dismissed OpenAI's appeal of a ruling that Ontario courts can hear the case.

The plaintiffs are a group of Canadian news publishers. It includes The Canadian Press, Torstar, The Globe and Mail, Postmedia and CBC/Radio-Canada. They sued in late 2024, alleging OpenAI copied their journalism without permission or payment to build ChatGPT.

The ruling settles where the case is heard, not who is right. OpenAI says its training is lawful. But for investors and corporate buyers, where a case is heard is not a small detail. OpenAI's argument was that a dispute over training AI belonged in American courts, where it relies on fair use, a defence that allows some copying without permission. That argument has now failed twice in Canada.

What the court decided

The appeal court upheld a 2025 decision by Justice Jessica Kimmel of the Ontario Superior Court. According to Broadcast Dialogue, a Canadian trade outlet, the three-judge panel was unanimous and Justice B.W. Miller wrote the ruling. Broadcast Dialogue reported that the court found Ontario has a "real and substantial connection" to the alleged breaches, the test Canadian courts use to decide whether they can hear a case.

The Canadian Press reported that the appeal court noted a gap in OpenAI's case: OpenAI did not file evidence that a Canadian judgment would conflict with US law. Broadcast Dialogue also reported that the court ordered OpenAI to pay C$100,000 toward the publishers' legal costs.

The lower court's reasoning is what makes the ruling matter beyond this one case. Justice Kimmel found that OpenAI could be treated as doing business in Ontario without an office there. Copyright lawyer Barry Sookman's summary of that decision lists the factors. They include a web crawler, software that automatically collects pages, gathering data from Canadian sites. They also include services, advertising and contracts aimed at Ontario users.

The lower court also found a strong enough argument that the publishers' website terms of use were binding contracts, and that scraping broke them. That means the case is not only about copyright. The publishers also claim breach of contract, unjust enrichment, and getting around technical measures meant to block copying, according to a case summary published by Lenczner Slaght, the law firm representing the publishers.

The 2025 ruling was not a complete win for the publishers. It let the case proceed against six OpenAI companies and stopped it against four others, according to Sookman's summary and a separate law-firm analysis of the decision.

Why it matters beyond OpenAI

The ruling gives a template to publishers outside the US. AI developers collect content from the whole web. If courts in each country where that content is published can hear claims under their own laws, AI companies face a patchwork of legal risk, not one fight at home.

Canada is already testing that. Postmedia, Torstar, The Globe and Mail and The Canadian Press filed a similar suit against Anthropic in Ontario in October, The Canadian Press reported. That claim alleges copying to build the Claude AI model, both for training and on an ongoing basis when users ask questions. Anthropic did not respond to the news agency's request for comment, and the claims have not been proven.

If Anthropic argues that suit belongs elsewhere, it will face an appeal ruling that has already rejected that argument from OpenAI.

The OpenAI case is one of dozens of AI-related cases in Canadian courts, including copyright and privacy claims, according to a tracker kept by the law firm Anand & Siu. They include a 2025 suit by Quebec's La Presse against OpenAI. In the OpenAI case, the publishers seek statutory and punitive damages and a permanent court order against the conduct, the tracker says.

What it does not settle

Nothing in Friday's ruling decides whether training AI on news articles breaks Canadian law. That fight comes later. OpenAI can still argue that Canada's Copyright Act does not reach copying done outside Canada. Sookman, a critic of parts of the lower court's reasoning, has noted those arguments remain open at trial.

OpenAI has said its models are trained on publicly available data and are grounded in fair use and related international principles, it told the Associated Press when the suit was filed. Initial coverage of Friday's ruling did not include a new comment from OpenAI. It did not say whether OpenAI would ask the Supreme Court of Canada to hear a further appeal.

Lawmakers have left the courts to decide. News and cultural groups have pressed Ottawa to set rules on how AI systems use copyrighted work. The Canadian Press reported that the government has not signalled plans to update the Copyright Act.

For business readers, the practical point is plain. Licensing deals with publishers are starting to look less like goodwill and more like insurance. Each country that lets local publishers sue at home adds legal cost and uncertainty for AI developers that have not paid for the content they use.