Publishers sue Cohere for copyright infringement over AI training
A consortium of 14 major publishers sued AI startup Cohere for alleged copyright infringement, claiming the company used over 4,000 copyrighted articles without permission to train its AI models and generated fake content attributed to the publishers. Cohere denied the allegations and called the lawsuit frivolous.
What happened
A consortium of 14 publishers, including Newsday, Condé Nast, Los Angeles Times, McClatchy Media Company, Politico, The Atlantic, Forbes, and Vox Media, filed a lawsuit against Cohere on February 13, 2025, in U.S. District Court in Manhattan, alleging the AI company used more than 4,000 copyrighted articles without permission to train its AI models and to generate content for users.
The publishers allege Cohere copied content behind paywalls and created fake content attributed to publishers. Cohere spokesperson Josh Gartner said the lawsuit was 'misguided and frivolous' and that the company 'strongly stands by its practices for responsibly training its enterprise AI.'
Reported impact
The publishers allege the unauthorized use harmed their brands and referral traffic by displaying large portions or entire articles to users without authorization and by attributing fabricated content to the publishers.
Index analysis
The lawsuit alleges Cohere engaged in 'massive, systematic' copyright infringement by scraping publisher content, including material behind paywalls, despite publishers' explicit attempts to block such scraping.
Governance controls
Cohere stated it has 'long prioritized controls that mitigate the risk of IP infringement and respect the rights of holders,' but said it would have welcomed a conversation with the publishers before learning of their concerns in the lawsuit.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.