AI Industry

Canadian Newspapers Sue Anthropic Over Claude Training Data

Oct 20, 2026·5 min read·By Rahish Saifi

Bundle of folded blank newsprint tied with jute twine on a weathered wooden loading dock, a single red maple leaf resting on top

A coalition of Canada's largest news publishers has sued Anthropic in Ontario, alleging the AI company used their journalism without permission to develop and operate Claude. The statement of claim was filed October 6, 2026 and drew wider attention after reporting on October 19. The allegations have not been proven in court, and Anthropic had not responded to requests for comment in coverage by CBC News, The Globe and Mail, and CTV News.

Key facts

Item	Detail
Plaintiffs	Postmedia, Torstar, The Globe and Mail, and The Canadian Press
Court	Ontario Superior Court of Justice (statement of claim dated Oct. 6, 2026, per Cartt.ca)
Defendant	Anthropic and a Canadian subsidiary (per Broadcast Dialogue)
Core allegation	Copyright infringement: publishers say Anthropic copied their content to build Claude and related services without a valid licence
Relief sought (per claim, as reported)	Permanent injunction; damages including punitive and statutory damages of up to $20,000 CAD per work infringed
Public reporting	First reported by Cartt.ca; national outlets published Oct. 19, 2026
Status	Allegations only — not adjudicated; Anthropic had not commented in initial reporting

Who is suing and where

The four plaintiffs are among the most prominent private news organizations in Canada. Together they say they produce a significant share of the country's journalistic output and reach millions of readers. Their lawsuit lands in Ontario, where Canadian media have already tested similar theories against another frontier-model vendor.

In 2024, the same publishers — alongside CBC/Radio-Canada — sued OpenAI over alleged use of news content. That case has not been settled, according to CBC's reporting on the Anthropic filing. The new action narrows the target to Anthropic and Claude, but the underlying grievance echoes the OpenAI fight: newsrooms argue that large language models are commercial products built on copyrighted reporting they were never paid for.

What the publishers allege

The statement of claim, as quoted by CBC and Cartt, frames Anthropic as a company that profited from Claude while bypassing publishers. Plaintiffs allege that "Anthropic has capitalized on the commercial success of its AI models, building an expansive suite of services, and raised significant capital — all without obtaining a valid licence from any of the News Media Companies." They further allege that "In doing so, Anthropic has been substantially and unjustly enriched to the detriment of the News Media Companies."

On data collection, the claim says Anthropic deliberately "scrapes" enormous volumes of text from the internet and from the plaintiffs' websites. The claim says: "To obtain content for ingestion into its models and services, Anthropic deliberately 'scrapes' (i.e. accesses and copies) enormous amounts of text data from the internet and from the Plaintiffs' Websites." They also allege Anthropic breached website terms of service and circumvented technological protection measures meant to restrict copying — such as paywalls, according to Cartt.ca, and network-level protections and login authentication, according to Broadcast Dialogue.

The publishers do not limit their complaint to one-time training crawls. They allege Anthropic continues to scrape in response to user prompts, and that Claude can output plaintiffs' articles verbatim, in paraphrase, or as summaries. Cartt reports that plaintiffs say those outputs can divert traffic, weaken advertising revenue, and undercut paid subscriptions. The claim also alleges Anthropic could choose licensing deals like other AI firms in other markets but instead, in the claim's words as quoted by Cartt, "Rather than seek to obtain the content legally, Anthropic has elected to brazenly misappropriate the News Media Companies' valuable intellectual property and convert it for its own uses, including commercial uses, without consent or consideration."

Plaintiffs further allege that news content is especially attractive for model training because of its quality and reliability, and that Anthropic controls which sites feed its systems. They cite a pattern of public criticism and other lawsuits, alleging "Notwithstanding concerns raised by several content creators across various industries, including through numerous other lawsuits and public criticism, Anthropic has done nothing to curtail its unlawful behaviour." Again, these are unproven allegations at this stage.

What they want from the court

According to Cartt's reporting on the claim, publishers seek a permanent injunction to stop the alleged infringement, plus damages. That includes punitive damages and statutory damages of up to $20,000 per work infringed under Canadian copyright law as pleaded — a figure that could compound quickly if a court were to count individual articles as separate works. Broadcast Dialogue reports the claim covers about 4.2 million works from The Canadian Press, adds breach of contract and unjust enrichment claims, seeks disgorgement of profits, and that no defence has been filed yet. No damages have been awarded; the case is at the pleading stage.

How this fits a wider wave of news-vs-AI cases

Canadian publishers are not alone. In the United States, The New York Times sued OpenAI over use of copyrighted material, and CBC notes the U.S. Justice Department sided with the tech company last month in that dispute — a signal that outcomes may differ by jurisdiction. Anthropic agreed in 2025 to pay $1.5 billion to settle a U.S. class action by authors over infringing activity, which Cartt describes as the largest settlement of its kind in U.S. history, approved in July. Claude became available to Canadians in June 2024, per Cartt.

The Canadian suits sit alongside Ottawa's unfinished policy debate. Publishers have pressed the federal government to take a clearer stand on AI and copyrighted news after the government's national AI strategy failed to address the issue, The Canadian Press reports. A single Ontario Superior Court case does not settle national copyright policy by itself, but it adds pressure on both legislators and AI vendors as Anthropic scales revenue and eyes public markets.

What it could mean for AI companies and Claude users

Analysis. If courts accept even part of the publishers' theory, AI vendors may face sharper incentives to negotiate Canadian licensing deals or to document opt-outs, robots.txt compliance, and training corpora exclusions. The allegation of ongoing scraping tied to live prompts blurs the line between training-time copying and retrieval-time reproduction — a distinction some defendants have relied on elsewhere. Users of Claude in Canada are not parties to the suit, but product changes — stricter citation rules, degraded answers on Canadian news topics, or geofenced retrieval — could show up if Anthropic adjusts risk posture.

Technical teams should watch how plaintiffs frame technological protection measures. If courts treat paywalls and rate limits as locks under copyright law, crawler policies and third-party dataset vendors become compliance surfaces, not just engineering choices. Anthropic has publicly invested in defenses against model distillation; this case tests a different boundary — whether inbound scraping of publisher sites is lawful in the first place.

What to watch next

Anthropic's statement of defence — and any public comment after the Oct. 19 news cycle — will define its legal and PR strategy. Watch whether the Ontario court consolidates or distinguishes this case from the 2024 OpenAI action, and whether Ottawa responds with copyright guidance or collective licensing proposals. Settlement talks could mirror the U.S. authors' deal, but statutory damages per work give publishers leverage that pure injunctive relief might not.