Newspapers already filed suit against OpenAI
By Ahmad Hathout
Canada’s major newspapers are alleging in a lawsuit filed against Anthropic that the company is scraping their content to train its artificial intelligence models without consent, according to a statement of claim filed in Ontario Superior Court of Justice this month.
The San Francisco-based company behind the family of large language models (LLMs) known as Claude is scraping text data from the news websites and then using that “proprietary content to develop its models, without authorization or consent,” says the October 6 statement of claim from Torstar, The Globe and Mail, Postmedia, and The Canadian Press.
In response to user inputs, the models will spit out part or all of the copyrighted content – either verbatim, paraphrased or in summarized form – which drives traffic away from the news websites, discourages advertisers to pay for placements, and reduces the need for paid subscriptions, the lawsuit claims.
“At all times, Anthropic was and is aware of its obligations to obtain a valid licence to use the Works,” the claim continues. “Other AI companies have engaged meaningfully with publishers in other jurisdictions, resulting in legitimate market-based licensing solutions that fairly compensate content creators. 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.”
The lawsuit notes that content from major news publications is preferred for ingestion for their higher quality and reliability compared to other internet sources. And the papers claim Anthropic has control over which sites are used to train its models.
The newspapers are claiming that Anthropic breached their terms of use, infringed on their copyrighted work, and have circumvented technological protection measures, such as paywalls. They are asking for a permanent injunction against the activity and damages, including punitive and the statutory $20,000 per work infringed.
Anthropic did not respond to a request for comment.
According to a U.S. lawsuit filed by music publishers, Anthropic’s founder and CEO Dario Amodei said, “[AI] models shouldn’t output copyrighted content because it’s—it’s against the law to do so.”
Citing internal communications in the company’s early days, however, a court noted in the high-profile proposed class action case of Bartz et al v. Anthropic that Amodei “prefered to steal” books the company was using to train its models to avoid the “legal/practice/business slog.”
While the company was found to have a fair-use claim to train its models on the digital conversion of purchased books, the U.S. District Court of the Northern District of California ruled it couldn’t make that claim for the over seven million books it downloaded from pirate sites, for which the class was approved for trial. The court also said the company could not just buy a copy of the books after the fact to absolve itself.
Later that summer, the company agreed to pay $1.5 billion to settle the matter for that infringing activity up to August 2025. The landmark settlement – the largest of its kind in U.S. history – was approved this July.
Created by former employees of ChatGPT maker OpenAI, Claude was made available to Canadians in June 2024 and generates revenue from subscription services and partnerships with major tech companies like Amazon, Microsoft and Google. The company has established deep roots on Canadian soil, including partnering with Telus and other institutions in the country.
According to a July post from Anthropic, Canada represents 2.6 per cent of global Claude.ai traffic and ranks 8th overall.
“Per capita, Canadians use Claude more than four times as much as would be expected given the size of their population,” the post says. “Among the ten countries with the highest Claude usage, only the United States surpasses it in this regard.”
That said, Anthropic is not the only AI company under the microscope and weight of lawsuits.
These same newspapers filed a lawsuit in November alleging OpenAI has been using their work to train its LLMs and reproduce it for commercial purposes.
The Society of Composers, Authors and Music Publishers of Canada (SOCAN) sued Suno, as others have done, for allegedly training its custom song AI models on the copyrighted works it represents. The suit came after the company said in an August blog post that, “AI should help people create something new, not imitate someone else’s work.”
In late July, a German court found Suno infringed copyright by training its AI model on six well-known songs of Germany’s preforming rights organization GEMA and allowing for their reproduction. GEMA also obtained judgment against ChatGPT maker OpenAI from the same court last year for using song lyrics without a licence when training its models.
Earlier this year, a trio of Canadian writers filed a proposed class action proceeding against Google in Federal Court for allegedly using unlicensed copyrighted works to train and develop its generative AI tools.
In the United States, news and magazine publishers – including Fox, The Atlantic, Forbes, The Guardian, The LA Times, Vox, and Torstar – filed a copyright infringement lawsuit against Canada’s Cohere, alleging the company’s models spit out infringing summaries of their work.
The issue of AI-generated news summaries has caught the attention of the federal government, which has inquired about an “opt-out” mechanism for Google’s AI Overview feature that generates summaries that news publishers say are driving down traffic to their sites.
Canadian Heritage has been aware of these lawsuits. A December memorandum to Culture Minister Marc Miller, obtained by Cartt, noted a “trend away from the use of copyright material by Artificial Intelligence training data to the issue of outputs.
“Recent developments indicate that companies could face exposure when they can produce outputs that too closely mirror protected expression,” the memo said.
The issue of AI outputs has also attracted Parliament’s attention. The Standing Committee on Canadian Heritage recommended copyright protections for news content scooped up by bots. Specifically, it said the federal government should require “clear opt-in consent requirement for the use of copyrighted works in the training of artificial intelligence systems, ensuring that creators’ works may not be used for text and data mining [TDM] or model development without their prior authorization.”
Photo via Flickr
