Canadian news outlets just sued OpenAI over copyright

The legal walls are closing in on OpenAI’s training playbook, and this time the challenge is coming from north of the border.

A heavyweight coalition of Canadian news publishers just filed a joint copyright infringement lawsuit against OpenAI in the Ontario Superior Court of Justice. The group includes CBC/Radio-Canada, Postmedia, Metroland, the Toronto Star, The Globe and Mail, and The Canadian Press.

They accuse OpenAI of capitalizing and profiting from their reporting to build products like ChatGPT without permission or compensation. Along with an injunction banning future scraping, the publishers are asking for punitive damages and a full disgorgement of profits OpenAI generated from using their articles.

No more piecemeal licensing

In a joint statement, the Canadian publishers pushed directly back against the tech industry’s favorite defense. They argued that journalism serves the public interest, but scraping rival content for private commercial gain does not.

OpenAI fired back with its standard defense, telling Engadget that training its models on publicly available online data is grounded in fair use and international copyright principles. The company also pointed to its content partnerships and its opt-out mechanism for publishers who do not want their sites indexed.

OpenAI also noted that its built-in ChatGPT search engine crawls the web simply to direct users to sources with attribution, rather than training models on that search-crawled data. Still, publishers are not buying the distinction between searching the web and vacuuming up the archives for foundation model training.

The pile-on continues

This new Canadian lawsuit adds to a mountain of legal trouble for the AI giant. OpenAI is already fighting high-stakes copyright battles with The New York Times, The Intercept, Raw Story, nonfiction authors, and comedian Sarah Silverman.

The optics for OpenAI keep getting trickier. Earlier this year, the company admitted in a submission to a UK committee that training frontier AI models without copyrighted materials would be impossible.

To make matters worse, The New York Times filed a declaration in court this month claiming that OpenAI engineers accidentally wiped out key training data evidence. A coordinated strike from Canada's largest media brands only adds fuel to the fire.

Why it matters

OpenAI has spent the past year striking bespoke licensing deals with select publishers, but this collective lawsuit proves that selective partnerships are not enough to avoid industry-wide fallout. By banding together across an entire national media landscape, the Canadian publishers are trying to force a systemic revenue-sharing model rather than accepting crumbs.

For users and developers, this will not break ChatGPT tomorrow, but the cost of building frontier AI is clearly shifting. If courts reject fair use arguments, model providers will face soaring licensing overheads, which will inevitably show up in API pricing and enterprise contracts.

We think OpenAI is trapped in a dilemma of its own making. Having already conceded to UK lawmakers that scraping copyrighted material is a structural necessity for frontier intelligence, defending that practice as harmless commercial fair use looks increasingly unsustainable in court.