A coalition of major news publishers has launched a legal battle against Cohere, alleging the AI startup has engaged in widespread copyright and trademark infringement. The lawsuit, filed in the Southern District of New York, claims that Cohere used thousands of copyrighted articles without permission to train its AI models, bypassing publishers’ websites and even generating false attributions to well-known news brands.
Key Points:
- The lawsuit accuses Cohere of training its AI on at least 4,000 copyrighted articles without authorization.
- Plaintiffs argue Cohere’s AI reproduces full or near-verbatim copies of their content, including paywalled material.
- Cohere is also accused of “hallucinating” articles and falsely attributing them to reputable publishers.
- The publishers seek damages of up to $150,000 per work and a court order forcing Cohere to destroy all copyrighted material in its possession.
According to the lawsuit, Cohere has allegedly used at least 4,000 copyrighted works without permission, reproducing portions—or entire copies—of articles in response to user queries. The plaintiffs argue this practice not only violates copyright law but also damages publishers’ ability to attract web traffic and subscription revenue.
Beyond copyright infringement, the publishers claim Cohere has misrepresented AI-generated content as their own. The lawsuit cites examples of Cohere “hallucinating” news articles—fabricating content and falsely attributing it to major publishers, an issue that could undermine the credibility of news organizations.
“This is an existential threat to journalism,” Condé Nast CEO Roger Lynch said in a statement. “Rather than create their own content, they’re stealing ours to compete with us without our permission, without compensation, and undermining our very business that feeds their machines in the first place.”
Cohere, a Canadian AI company valued at over $5 billion, denied the allegations. “We strongly stand by our practices for responsibly training our enterprise AI,” said Josh Gartner, the company’s head of communications. “We would have welcomed a conversation about their specific concerns rather than learning about them in a filing. We believe this lawsuit is misguided and frivolous and expect this matter to be resolved in our favor.”
The lawsuit is the latest in a string of legal challenges AI firms are facing over the use of copyrighted material. In 2023, The New York Times sued OpenAI and Microsoft for similar alleged violations, and News Corp. has taken legal action against AI-powered search engine Perplexity. While some news organizations have struck licensing deals with AI companies, others are choosing litigation to protect their intellectual property.
The plaintiffs are seeking statutory damages of up to $150,000 per work infringed, as well as an injunction to prevent Cohere from using their content. They are also demanding the destruction of any copyrighted material in Cohere’s training data. If the case proceeds, it could set a major precedent for how AI companies use copyrighted materials in training large language models.