Cohere was purpose-built for enterprise. We build models tailored for secure, business use cases — tools that help organizations search proprietary data, translate their documents, or monitor internal operations – so that companies can work smarter, faster, and better. There’s a huge market for AI chatbots that help consumers with travel recommendations, homework help, product reviews, and a million other things, but that’s not what we do.

Earlier this year, a group of media publishers sued Cohere in federal court, alleging copyright and trademark infringement. Today, we respond to several of their claims, outlining why the case is misguided and the assertions made show a lack of understanding about Cohere’s business and the uses of its technology. The Playground that the publishers highlight is not a Cohere product, but a developer-focused demo environment with clear terms of use and a narrow purpose. The ways that they used it bear no semblance to the reality of how these products are actually used in commercial applications.

Our technology helps companies in industries like finance, government, healthcare, technology, and manufacturing analyze their own data, identify patterns, create summaries, and similar tasks. We built it to facilitate, automate and optimize workflows in the customer’s own environment, not as a substitute for reading news articles from the web. And that is how and why our customers use it – not as just another of the many consumer-focused AI products out there.

We respect the critical role that media organizations play in society. But we strongly disagree with any suggestion that Cohere’s AI solutions undermine the media industry or its business model, and we look forward to demonstrating that our technology and user base respect intellectual property laws.

A Case Based on Manufactured Data

The key issue in the output-related claims of this litigation is that the group of publishers involved used a Cohere demo tool for developers (called “The Playground”) in a way purposely engineered to support their claims. All of the examples included are apparently from the plaintiffs themselves, and not from an actual Cohere user. There are no allegations in the plaintiffs’ complaint that an actual Cohere user has ever prompted any Cohere tool in the way this lawsuit alleges.

Our models generate summaries and other outputs based on the data sources our customers connect models to for their business purposes. To be clear, we have built these models to analyze your own internal data, which is why we say that AI is only as good as the data it can securely access: your data for your own use. We make a web search tool available to demo the search and retrieval capabilities of our solutions in enterprise use-cases, whose aims and objectives are altogether different. The plaintiffs misused and manipulated a developer-focused demo search tool to generate summaries and other outputs based on their own works.

The plaintiffs are intentionally constructing a scenario that would not occur from any real-world use case to form their argument. This is a foundational failing that dooms most of the plaintiff’s claims, and it’s for this reason that we are confident in our motion to dismiss, which was filed in U.S. District Court for the Southern District of New York today.

Cohere will always remain committed to transparency, responsibility and building AI that works for business, society and for the law.

Cohere files motion to dismiss in Advance Local Media suit