Merriam-Webster is suing OpenAI over ChatGPT training - qz.com

Encyclopedia Britannica and its Merriam-Webster subsidiary have filed a copyright and trademark lawsuit against OpenAI in federal court, alleging the company systematically scraped nearly 100,000 of its articles to train ChatGPT — and continues to reproduce that content verbatim in its outputs. The complaint, filed March 13, 2026 in the U.S. District Court for the Southern District of New York (case no. 1:2026cv02097), is the latest salvo in an escalating legal war between publishers and AI companies over who owns the data that powers large language models.

Britannica stopped printing its 32-volume encyclopedia in 2012. Today its business runs entirely on subscriptions and web traffic — which makes the alleged free ride by ChatGPT more than a legal nuisance. It is, the complaint argues, an existential threat to a publisher that has been writing reference articles since Edinburgh in 1768.

"ChatGPT then provides narrative responses to user queries that often contain verbatim or near-verbatim reproductions, summaries, or abridgements of original content, including plaintiffs' copyrighted works," the complaint states. The lawsuit frames this as three distinct violations: scraping Britannica's sites to build training inputs, feeding that content into GPT during training, and then generating outputs that reproduce or closely mirror the originals when users ask questions Britannica's catalogue covers.

The second pillar of the suit is trademark. By pairing AI-generated responses — which may include hallucinations or fabrications — with Britannica's and Merriam-Webster's brand identities, the complaint alleges OpenAI misleads users into believing the publishers have endorsed or sourced those outputs. Britannica's reputation, built over more than 250 years on accuracy, is at risk every time ChatGPT produces a confident error and slaps the Britannica name on it.

The filing copies the legal strategy Britannica used when it sued Perplexity AI in September 2025 over nearly identical allegations. That case is still ongoing. But the OpenAI matter lands in a more complex landscape: OpenAI is already the subject of a multidistrict litigation in the SDNY, overseen by Judge Sidney Stein, that consolidates dozens of publisher and author suits. Britannica's case joins that consolidation.

The precedent is unsettled. In September 2025, Anthropic convinced a federal judge that using content as training data is transformative enough to be legal — but Judge William Alsup still found Anthropic violated the law by downloading millions of books without paying for them, resulting in a $1.5 billion class action settlement. Britannica's case will test whether the fair-use argument holds when the plaintiff has a direct economic grievance rooted in their content being reproduced in outputs, not just used during training.

OpenAI's response, per a spokesperson: "Our models empower innovation, and are trained on publicly available data and grounded in fair use."

Britannica is seeking unspecified monetary damages and a court order blocking the alleged infringement. Given the MDL context and the stakes for the entire AI-publisher relationship, this one is worth watching closely.

Sources: Reuters — Encyclopedia Britannica sues OpenAI over AI training | The Next Web — Britannica and Merriam-Webster sue OpenAI | TechCrunch — The dictionary sues OpenAI | Complaint PDF | Perplexity lawsuit — The Verge | Anthropic $1.5B settlement — TechCrunch