Canadian AI company Cohere sued by major publishers for copyright violations

Globe and Mail

13-02-2025

More than a dozen publishers including Condé Nast, The Atlantic and the Toronto Star are suing Canadian artificial intelligence company Cohere Inc. for copyright and trademark infringement, alleging the company improperly scraped news content on a 'massive' scale to build its AI models.
The lawsuit, filed Thursday in New York, is the latest salvo between companies that produce content and tech developers that mine huge amounts of data from the internet to build AI software that can then generate text, pictures, video and audio.
Vox Media, Politico, Forbes and others are alleging that Cohere has used their content without consent or payment to build its technology, which can then reproduce those very same articles verbatim in some cases, even breaking news pieces and those behind paywalls. Cohere's models can also produce fake articles attributed to the news organizations, damaging their brands, according to the lawsuit.
'Cohere copies, uses, and disseminates publishers' news and magazine articles to build and deliver a commercial service that mimics, undercuts, and competes with lawful sources for their articles,' the court filing reads. 'Left unfettered, such misconduct threatens the continued availability of the valuable news, magazine, and media content that publishers produce.'
The publishers allege that they will suffer lower subscription and advertising revenue as a result. The lawsuit asks for monetary damages and an order requiring Cohere to destroy the copyrighted works in its possession, among other measures.
None of the allegations have been proven in court.
'We believe this lawsuit is misguided and frivolous, and expect this matter to be resolved in our favour,' said Cohere spokesperson Josh Gartner, adding that the company has prioritized tools to mitigate the risk of infringement. 'We would have welcomed a conversation about their specific concerns – and the opportunity to explain our enterprise-focused approach – rather than learning about them in a filing.'
The Toronto Star is the sole Canadian plaintiff in the lawsuit. 'Our business model does not work if the valuable content we produce is stolen and reproduced, without compensation, by AI companies,' wrote Neil Oliver, chief executive of parent company Torstar Corp., in an internal memo to employees. 'Nor will our industry maintain its sustainability if these companies are allowed to continue building their success at our expense.'
In November, the Toronto Star joined other Canadian news organizations, including The Globe and Mail and the CBC, in filing a similar lawsuit against OpenAI.
Cohere was founded in Toronto in 2019, and has since opened offices in New York, San Francisco and London. In July, it raised US$500-million at a US$5.5-billion valuation, and its investors include Radical Ventures, Oracle Corp. and Nvidia Corp. The company has differentiated itself from competitors such as OpenAI and Anthropic by developing AI models and tools for corporations, rather than the general public.
Last year, Cohere said it would indemnify customers who are sued for intellectual property violations and assume responsibilities for any legal settlements or judgments.
The statement of claim filed Thursday contains numerous examples of Cohere's AI models allegedly reproducing or paraphrasing news articles through the company's chatbot interface, which can be used for free. Asked to provide a specific article from the Los Angeles Times, Cohere's chatbot obliged with passages of verbatim prose, according to the example in the lawsuit.
In another example, Cohere's chatbot is asked what the Toronto Star reported about Ticketmaster stopping ticket transfers for Taylor Swift's Eras Tour owing to theft concerns. The answer allegedly included verbatim statements from a Toronto Star article and includes similar phrasing and organizational structure, 'signalling to the reader that the original text need not be consulted.'
A number of news organizations, including some of the plaintiffs in the lawsuit, have signed deals with other AI companies to use content in exchange for payment. 'Cohere fails to license the content it uses,' the lawsuit alleges, which gives the company an unfair advantage against AI developers that do strike agreements.
Publishers have taken steps to stop AI companies from using their material, such as code on their websites that effectively tells the software that mines data not to do so. 'Cohere nonetheless copies articles from these sites,' the statement of claim alleges.
The copyright regimes in the United States and Canada were ill-prepared for the generative AI boom, which kicked off with the release of OpenAI's ChatGPT in late 2022. AI companies have argued that their practices are legal, while publishers, artists and other content creators have sued and turned to the courts for answers.
The Canadian government has conducted a consultation about copyright and generative AI. Cohere, along with other AI developers, said during that consultation that it favours an exemption in copyright law that would allow it to build commercial models without being compelled to pay or obtain permission from content creators to use their material, warning that a requirement would hobble Canada's AI industry.