Skip to content

Signals/ Claude & Anthropic/ Brief /

Canadian news publishers sue Anthropic over Claude's training

Four of Canada's biggest news publishers say Claude was built on their articles without a licence. None of it is proven yet.

Postmedia, Torstar, The Globe and Mail and The Canadian Press have sued Anthropic, the company behind Claude, in an Ontario court. Reports say the claim was filed on October 6 and became public on October 19.

The publishers say Anthropic copied their articles to build Claude and had no licence to do so. They also allege it broke their websites’ terms of service, got around technical protections meant to stop copying, and keeps scraping their sites when users send prompts. The claim says Anthropic was “substantially and unjustly enriched” at their expense.

These are the publishers’ allegations. A court has not tested any of them. Anthropic had not replied to requests for comment in the coverage we read.

The same publishers, together with CBC/Radio-Canada, sued OpenAI in 2024, and that case has not settled.

The catch: The reports we read do not say what damages the publishers want, and the claim is only an accusation so far.

Background

AI companies train models on huge piles of text, and much of it comes from the open web. Publishers say that includes paid journalism copied without permission. The Canadian claim adds an allegation about scraping that happens when a user sends a prompt.

What to watch next

  • Whether Anthropic files a public response or defence.
  • Whether the OpenAI case from 2024 produces a ruling or settlement.
  • Whether other publishers outside Canada file similar claims.

Sources & links 2

  1. CP24: News publishers launch copyright suit against AI company Anthropic in Ontario cp24.com ↗
  2. MobileSyrup: Canadian publishers launch copyright suit against AI company Anthropic mobilesyrup.com ↗

More on Claude AI News, AI Safety & Regulation News

More signals

Reel

Claude Now Hides a Watermark in the Words It Writes. Here's Who Can Check It

Anthropic says text from new Claude models now carries an invisible watermark, worldwide. It isn't hidden characters: a secret key shapes Claude's word choices into a pattern a detector can spot. Here's how it works, where it's on, who can apply for the detector, what it means for students, writers and businesses, and the catch.

Helps Students, freelance writers, marketers and businesses who paste Claude's text into their work and want to know what the watermark does and doesn't reveal

  • Claude
  • Anthropic
  • AI Safety