We may earn a commission when you buy through links on this page. It doesn't change what we recommend.
A group of Canada’s largest news publishers has filed a lawsuit against Anthropic in an Ontario court, accusing the AI company of using their journalism without permission to train and run its Claude models. The claim, dated October 6 and widely reported this week, is the second time these publishers have taken an AI developer to court, and the Anthropic copyright lawsuit adds a new chapter to a fight that is being argued in courts on both sides of the border. The allegations have not been tested in court. This article covers who is suing, what they claim, what they want, and what the case could mean.
Who is suing and what they say
According to the Canadian Press, the plaintiffs include Postmedia, Torstar, The Globe and Mail, and The Canadian Press itself. Broadcast Dialogue, which read the filing, adds that the Toronto Star, Metroland Media, and PNI Maritimes are named, and that the defendants are Anthropic and its Canadian subsidiary. The suit was filed in Ontario Superior Court on October 6 and was first reported by CARTT.ca.
The plaintiffs describe themselves as producers of a significant share of Canada’s journalism, from print broadsheets to round-the-clock digital coverage, reaching millions of readers. That reach is part of their argument about harm.
The publishers say Anthropic copied their content to develop Claude without a valid license from any of them. They also allege the company ignored website terms of service, got around technical protection measures such as paywalls and logins, and keeps scraping text from their sites in response to user prompts. Toronto Life quotes the filing as saying Anthropic “has done nothing to curtail its unlawful behaviour.”
The claims in the Anthropic copyright lawsuit

The filing lists several causes of action. Based on the reports, they are copyright infringement, unjust enrichment, breach of contract through the sites’ terms of use, and circumvention of technological protection measures. Broadcast Dialogue says the publishers also argue that the conduct harms their market by diverting readers, reducing subscribers, and deterring advertisers.
Scale is part of the pitch. A table reported by Broadcast Dialogue lists works from 2021 onward: about 4.2 million for The Canadian Press, 3.5 million for Postmedia and PNI Maritimes, 3.5 million for Torstar, and 2.3 million for The Globe and Mail. Those figures add up to roughly 13.5 million articles, some owned and some licensed. Those are the publishers’ own counts, and the court would have to test them.
What the publishers want
The Canadian Press and Toronto Life reports don’t say what remedies are sought, but Broadcast Dialogue does. The relief includes damages for infringement, breach of contract, and unjust enrichment, an accounting and disgorgement of profits, punitive damages, and permanent injunctions barring Anthropic from accessing or using the publishers’ works without written consent. As an alternative, the claim asks for statutory damages of up to $20,000 for each work infringed.
A simple calculation shows why that number draws attention. If every one of roughly 13.5 million works were awarded the maximum, the total would be about $270 billion. No court is likely to go near that. A judge would have discretion to award far less, so the multiplication is a ceiling, not a forecast. It does explain why the publishers are leading with it.
The background

The same publishers, along with CBC/Radio-Canada, sued OpenAI in 2024, and that case is still unresolved, the Canadian Press notes. The new claim also points to the United States. It cites a California federal court finding that Anthropic had downloaded more than seven million pirated books, a dispute that ended in a US$1.5 billion settlement, according to Broadcast Dialogue’s account of the filing. The publishers contrast Anthropic with AI developers that have signed licensing deals with media companies, saying Anthropic has no public licensing agreement with major publishers.
The claim also cites Anthropic’s size. It says the company had raised about US$125 billion and was valued above US$965 billion as of May, and presents that as context for the demand for payment. We haven’t independently verified those figures.
What Anthropic has said
Anthropic hasn’t responded to requests for comment, the Canadian Press reports, and Broadcast Dialogue says no statement of defence has been filed. Under Ontario rules, defendants served outside Canada can have up to 60 days to respond. Until Anthropic files, we only have one side’s description of events. Early steps in civil cases like this usually involve pleadings and procedural motions, so a trial is likely far off, and a settlement or licensing deal could also end the dispute before any ruling.
Why the Anthropic copyright lawsuit matters beyond the courtroom
For newsrooms, the practical question is whether AI companies must license journalism. A ruling in the publishers’ favor could push AI developers toward paid agreements, while a ruling for Anthropic could slow those talks. Another AI business shaped by rights-holders is Character.AI, which we covered in our report on its bankruptcy and the Disney bid. For companies that use AI tools, the case is one more example of legal risk that some insurers are moving to exclude, as we described in our coverage of AI exclusions in liability policies.
What to watch next
Several things are worth watching in the Anthropic copyright lawsuit. First, Anthropic’s statement of defence and whether it challenges Ontario as the forum. Second, any movement in the OpenAI case, which began in 2024. Third, whether Canada’s federal government responds to the publishers’ request for a clear position on AI and copyright, since they say its national AI strategy was silent on the issue. A final possibility, though nothing in the filings suggests it is close, is a licensing deal between the two sides, which would change the story entirely. This article is general information, not legal advice.
Editor’s take
The Anthropic copyright lawsuit will be slow, and the headline numbers are leverage, not predictions. What matters is the question underneath: who pays for the journalism that AI systems learn from? The Canadian filing is a long way from an answer, and the case may take years. The evidence will come from the court record, not from either side’s press statements.








