Britannica and Merriam-Webster Sue OpenAI Over AI Training

Two of the world’s most trusted reference publishers, Encyclopedia Britannica and Merriam-Webster, have initiated a major copyright and trademark lawsuit against OpenAI. The legal action, filed in a New York federal court, alleges that OpenAI’s ChatGPT was trained on and systematically reproduces their copyrighted content without permission, causing significant commercial and reputational harm. This case represents a critical escalation in the ongoing legal battles over how artificial intelligence models are trained and whether they unfairly exploit copyrighted material.

The complaint details that OpenAI used nearly 100,000 of Britannica’s online articles as training data for its language models. It argues that ChatGPT’s responses to user queries frequently contain verbatim excerpts, close summaries, or condensed versions of the publishers’ original works. This, the plaintiffs state, constitutes direct copyright infringement at multiple stages: the initial scraping of their websites, the use of that content in model training, and the subsequent generation of outputs that replicate the protected material.

A second major claim revolves around trademark law. The lawsuit contends that by generating responses, which can include inaccuracies or “hallucinations”, in proximity to the famous Britannica and Merriam-Webster brands, OpenAI misleads users. This creates a false impression that the publishers endorse or are the source of the AI’s answers, thereby inflicting reputational damage on institutions built over centuries on a foundation of accuracy.

The core business argument is one of economic survival. Britannica, which long ago transitioned from print to a digital subscription and advertising model, asserts that ChatGPT diverts web traffic and undermines its revenue streams. When users get answers sourced from Britannica’s editorial catalogue directly from the AI, they have no need to visit the publisher’s websites. The lawsuit frames this as OpenAI taking a “free ride” on high-quality, trusted content, effectively transferring the value created by Britannica’s researchers and editors to its own platform without compensation.

This legal action follows a nearly identical lawsuit the same plaintiffs filed against the AI search engine Perplexity in late 2025. That case, which also accuses Perplexity of scraping protected content and presenting it as AI-generated summaries, remains ongoing. The case against OpenAI, however, enters a far more complex legal environment. It joins a large consolidated proceeding in New York that already encompasses over a dozen copyright lawsuits from major news publishers against OpenAI. Legal analysts predict the Britannica case will likely be folded into this broader litigation, potentially delaying a final resolution for years.

The filing arrives amid a stark divergence in how content creators are responding to AI. While many publishers are pursuing litigation, others are opting for licensing agreements. Significant deals, like one between News Corp and Meta, demonstrate an alternative path. The legal landscape is already showing financial stakes, as evidenced by a $1.5 billion class-action settlement in a separate case involving Anthropic and pirated book data.

Whether this lawsuit ends in a settlement, a trial, or prolonged consolidation remains uncertain. What is clear is that publishers with formidable brand authority and a vested interest in factual integrity are choosing to fight. They are signaling an unwillingness to stand by while AI systems they believe are built on their work reshape the market for reliable information without their consent or fair compensation.

(Source: The Next Web)