A consortium of 14 publishers including Conde Nast, Forbes and The Atlantic are suing AI startup Cohere for copyright and trademark infringement, accelerating the news industry’s legal battle with technology. The news publishers are accusing the Canadian company, valued at $5 Billion of illegally using more than 4000 of their copyrighted works to train its AI chatbots. The suit alleges that Cohere engages in widespread unauthorized use of publisher’s content in training their AI generative systems as well as display entire articles without directing the reference traffic to the publisher’s website. This amounts to massive systematic copyright as well as trademark infringement.
Additionally, Cohere is also accused of generating “hallucinated content” and factually inaccurate information that wasn’t actually published on the publisher’s website. Rather than answering the user queries and directing them to the publisher’s page, the Cohere’s AI technology provides the users with a full, nearly verbatim text from an article by altering merely 2 lines, including from the articles that are kept behind the paywalls. The lawsuit is filed in the Southern District of New York and the complainants seeks damages of maximum amount of $150,000 per infringed work under the Copyright Act, permanent injunction as well as an order to completely destroy all the copyrighted works in Cohere’s possession.
Cohere denies all the allegations against them, claiming that their practices are in tune with the Intellectual Property Rights. In a public statement, Josh Gartner, head of communications at Cohere, states that Cohere strongly stands by its practices of responsibly training AI and would’ve welcomed the conversation about the publisher’s concerns and an opportunity to explain their part of enterprise-focused approach rather than learning about them in a lawsuit. He terms the present suit as “misguided” & “frivolous”.
The current law suit piles onto the prevailing tensions between the publishers and the AI companies that use news and other content to train their AI models. Publishers state that the AI chatbots threaten their businesses which largely rely on revenue generated by advertising and subscriptions. The ultimate goal of the present suit is to establish a legal precedent and formulate terms for playing the field for licensed use of journalism by AI, by allowing them to use their content for training AI models and real-time uses in exchange of payments.
The lawsuit against Cohere is latest one in the string of lawsuits filed against AI companies for alleged IP violations. Similar issues have surfaced internationally. In December 2023, The New York Times sued OpenAI and their large financial backing Microsoft. The Wall Street Journal and New York Post similarly sued perplexity in October over issues regarding unauthorized use of their copyrighted content by AI companies for training their AI models that provides comprehensive answers to its users on a single prompt. In a similar situation in India, the digital news companies of Indian billionaires Mukesh Ambani (News 18 Network) and Gautam Adani (NDTV) along with several other media organizations filed a lawsuit against OpenAI, accusing them of using their copyrighted content without prior authorization. It accuses the San Francisco based AI giant of “wilful scraping” and wrongfully exploiting the publisher’s works to generate revenue while leaving the original content creators without any financial benefits. Courts across the worlds are hearing claims made by the authors, media organizations, musicians and artists suing the AI- tech firms of using their copyrighted content to train AI chatbots and formulating a well-thought approach towards the issue.
The argument of AI vs Copyright is a widely debated issue that lies on the intersection of Intellectual Property Law, Law of fair use and emerging technological advancements. Current legal frameworks provide some guidance on the use of copyrighted material to train AI models however these frameworks may not fully address the complexities of the AI generative systems. Balancing technological advances along with the preservation of creative rights of the authors is critical while navigating the difficult landscape of copyright and AI. Moving forward, several media companies including the Plaintiffs in the present case, have signed licensing agreements with AI companies like OpenAI, mutually allowing them to use their content in exchange for payment. Rather than stealing the original content without prior permission or paying compensation and undermining the very source that feeds the machines in the first place, it is suitable to execute a license agreement which mutually benefits both the parties. From IP Law’s perspective, this issue is far from settled. Courts will play a crucial role in determining whether AI training practices comply with the existing copyright frameworks. Moreover, formation of clearer AI- specific copyright frameworks are required to balance innovation and creator rights. This area of law is the one to keep our eyes on as it could shape future regulations and legal careers in AI, IP and Technology Law.
Authors: Seema Meena, Manasvi Shah & Devanshi Gotecha