Canada's news publishers are taking the AI training fight to court at home. A coalition that includes Postmedia and The Globe and Mail has filed a lawsuit in Ontario alleging that Anthropic used publishers' content to train Claude, its family of artificial intelligence models, without permission, according to a roundup of Tuesday's AI developments by The Neuron.
The claims remain allegations before the court; they have not been proven, and Anthropic's response in the proceeding will set out its position. Litigation of this kind typically turns on whether training on published material without a licence infringes copyright, or falls within exceptions that AI companies argue cover the automated analysis of publicly available content.
The Ontario filing extends a global pattern. News organizations in several countries have sued or struck licensing deals with AI developers over training data, arguing that models built on journalism should pay for the journalism. Publishers also point to a second pressure: AI answer engines that summarize news can reduce the traffic that funds newsrooms, an issue Canadian outlets have raised since Ottawa's battles with platform companies over news compensation.
For Anthropic, the case adds a Canadian front to its legal exposure at a moment when the company is also reportedly in talks over a further $45 billion compute deal with Nscale and expanding Claude into mobile and web coworking tools. The outcome in Ontario could influence how AI firms license Canadian news archives — and how much of the AI economy's value flows back to the newsrooms whose work helps train it.



