Britannica and Merriam-Webster sue OpenAI over alleged copying

In short: Encyclopedia Britannica and Merriam-Webster sued OpenAI in New York federal court, saying it copied their copyrighted work to train its AI models without permission or payment.

What happened

Encyclopedia Britannica and Merriam-Webster filed the lawsuit on March 13, 2026. They say OpenAI systematically took content from their websites and used it to train large language models, which are AI systems that learn patterns from huge amounts of text (like a very powerful autocomplete).

The complaint says the copying covered almost 100,000 articles and other reference material. It claims OpenAI used the content to build models including GPT-4 and later versions.

The lawsuit also mentions retrieval-augmented generation, or RAG, which is a way for an AI to look up text from a collection of documents while answering (like checking notes while speaking). The publishers argue that OpenAI copied not just small facts, but longer explanatory passages that took years and significant money to create.

The court documents also include examples meant to show close matches. One example described involves a user asking an AI search tool to define “plagiarize,” and the answer allegedly mirroring Merriam-Webster’s definition, including usage examples, without clear credit.

Beyond copyright, the publishers also claim trademark harm. They argue that if an AI wrongly attributes incorrect information to Britannica or Merriam-Webster, it could damage their reputations for accuracy.

The lawsuit further alleges “stealth crawling,” meaning OpenAI bypassed website rules such as robots.txt files, which are like “do not enter” signs for automated web scrapers.

Why it matters

This case is part of a bigger fight over who should get paid when AI systems learn from books and websites. If the publishers win, it could change what information AI tools can use, and how often users see paywalls, licensing deals, or stricter limits on automated copying.

Source: TechCrunch AI