Ontario Court of Appeal Lets Copyright Suit Against OpenAI Proceed

Ontario's Court of Appeal on Friday dismissed OpenAI's bid to shut down a copyright lawsuit by Canadian news publishers, clearing the first Canadian case over generative AI training on copyrighted news content to move toward a trial on the merits. The court also ordered OpenAI to pay $100,000 in legal costs.

The unanimous decision, written by Justice Bradley Miller, upholds a ruling from Superior Court Justice Jessica Kimmel, who found last year that three OpenAI-affiliated entities — OpenAI OpCo, LLC, OpenAI, LLC and OpenAI, Inc. — were "doing business" in Ontario. The coalition of publishers, which includes the Toronto Star, alleges OpenAI illegally used their articles to train ChatGPT and is profiting unjustly from that use.

What did OpenAI argue, and why did it fail?

OpenAI, based in California, argued that because it has no physical office in Ontario, the dispute belongs in a U.S. court. Kimmel rejected that argument, and Miller wrote that she did not err in finding multiple instances of the companies carrying on business in the province. Those instances include:

• Contractual arrangements
• Advertising to Ontario users
• Licensing tools to customers directly or through third parties

OpenAI had "relied heavily" on the absence of a physical presence, Miller noted, but the case law it cited did not support the proposition that physical presence is required for an Ontario court to assume jurisdiction. The company did not respond to a request for comment by publication time.

Who stands behind the case, and what are they saying?

Sana Halwani, a lawyer at Lenczner Slaght LLP representing the publishers, framed the stakes broadly: "This decision matters well beyond the news media industry. Digital businesses earn money from Canadian content and users, and their activity reaches into Canada. When disputes arise, they must answer to Canadian law in Canadian courts." Her firm added that the media companies welcome technological innovation but insist all participants follow the law.

Alissa Centivany, an associate professor of information and media studies at Western University, said the outcome could affect "a big range of copyright rights holders in Canada." Most jurisdictions permit scraping for non-commercial or research purposes, she noted, "but we don't have any jurisdictions that have said it's OK to do this for commercial reasons or for reasons that involve creating expressive outputs that compete with the originals." She added that she hopes courts will find such use infringing.

Could this still reach the Supreme Court of Canada?

OpenAI's options are not exhausted. The company may still seek leave to appeal to the Supreme Court of Canada — though that court hears only a small fraction of the cases brought before it, so the odds of a further round are long.

The ruling lands amid a wave of parallel litigation. The New York Times sued OpenAI in 2023, and eleven other U.S. publishers have since joined that case. In Canada, the same coalition of news organizations has sued Anthropic over alleged copyright infringement in training its Claude models, and La Presse has filed a separate suit against Anthropic in Quebec. Class actions on behalf of Canadian authors are pending in British Columbia and Quebec.

For now, the Ontario case returns to the lower court, where the publishers' claims of copyright infringement and unjust enrichment will be tested against whatever defences OpenAI mounts. Friday's decision establishes only one thing: a Canadian court gets to hear them.