Encyclopaedia Britannica and Merriam‑Webster have filed a federal lawsuit in the U.S. District Court for the Southern District of New York alleging that OpenAI used their copyrighted reference works without permission to train its ChatGPT language models. The complaint asserts that nearly 100,000 Britannica online articles and Merriam‑Webster dictionary entries were scraped and included in OpenAI’s training data, that the models sometimes reproduce or closely summarize Britannica content in whole or in part, and that AI-generated answers citing Britannica can contain fabricated information. The plaintiffs say those outputs divert readers from their websites, harming subscription and advertising revenue, and that instances where the models attribute content to Britannica or imply publisher endorsement give rise to trademark claims under the Lanham Act.

Britannica and Merriam‑Webster seek monetary damages, a court order to halt the alleged copying and further use of their material in OpenAI’s systems, and stronger protections to prevent AI systems from reproducing copyrighted works. The complaint also challenges OpenAI’s alleged use of Britannica content within retrieval‑augmented generation features.

OpenAI has rejected the claims, stating its models are trained on publicly available data, that its systems transform information rather than simply copy it, and that fair use principles apply; in one summary OpenAI did not comment before publication. The filing joins a broader wave of lawsuits by news organizations, authors, music publishers and other content creators against AI companies, including related suits by The New York Times, Ziff Davis, more than a dozen newspapers, and a prior Britannica suit against Perplexity. A separate multidistrict litigation in the same court consolidates more than a dozen copyright suits by news publishers; the Britannica filing is likely to be transferred into that MDL and paused pending its outcome.

Legal observers quoted in coverage say these cases could affect how AI companies source training data, whether licensing deals will be required, and how courts define fair use for AI training. The legal landscape remains unsettled: a federal judge in a different case has described training use as potentially transformative while still finding unlawful downloading that warranted a class action settlement. The outcome of the Britannica and Merriam‑Webster lawsuit may therefore influence the data sources, construction, and commercial relationships surrounding large language models and the amount of web traffic directed to original content providers.

Original Sources: 1, 2, 3, 4, 5, 6, 7, 8 (openai) (authors)

No real value analysis available for this item

No bias analysis available for this item

The text expresses several emotions, some explicit and some implied, that shape how a reader understands the dispute. Concern and alarm are present in phrases describing the lawsuit and its stakes: words such as “alleging,” “without permission,” “halt the alleged misuse,” and “stronger protections” convey worry about rights being violated and potential harm to publishers. This concern is moderate to strong because it frames the publishers as harmed parties seeking legal remedy and protection, giving the complaint urgency and seriousness. The effect of this worry is to invite the reader’s sympathy for Encyclopaedia Britannica and Merriam-Webster and to raise doubts about the practices of the AI company. Defensive confidence and denial appear in OpenAI’s response, which “rejected the claims,” says models were trained on “publicly available data,” and invokes “fair use.” These phrases show a calm, assertive stance intended to reassure readers and lessen the perceived wrongdoing. The strength of this emotion is measured and purposeful: it aims to reassure regulators, customers, and the public that the company acted lawfully, steering readers toward trust in OpenAI’s position. Tension and anticipation appear through references to a “broader wave of legal actions” and the statement that “the outcome of the case could affect” how models are built and traffic patterns. This creates a sense of looming consequence and uncertainty that is moderate in intensity; it signals that the case has wide implications and keeps readers alert to future developments. The likely effect is to engage interest and seriousness about the broader industry impact. Accusation and indignation are implicit in the publishers’ claim that models “can reproduce content…in near-verbatim form” and that AI answers are “diverting traffic away” from original sites. These choices of wording heighten the sense of unfairness and loss and are moderately strong, encouraging readers to view the publishers as injured and to be critical of the technology’s effects on content creators. Finally, impartiality and analysis appear through phrases like “legal experts quoted” and “could influence,” which introduce a neutral, analytical tone that tempers emotion with consideration of legal and practical consequences. This moderating tone is mild but important: it guides readers to see the situation not only as a dispute but as a matter with policy and industry implications, encouraging thoughtful attention rather than purely emotional reaction. The emotions steer reader response by building sympathy for the publishers, offering reassurance from OpenAI, generating concern about broader consequences, and prompting critical thought about industry practices. The writer uses emotive word choices—“alleging,” “without permission,” “rejected the claims,” “diverting traffic,” and “could influence”—to make stakes and positions clear rather than neutral. Repetition of the dispute’s scope (nearly 100,000 articles, encyclopedia entries, dictionary definitions) amplifies the scale and gravity of the claim. Contrasting language—publishers’ demands for damages and protections versus OpenAI’s fair use defense—creates a clear adversarial frame that increases tension. Mentioning other lawsuits and “legal experts” broadens the context, making the issue seem systemic rather than isolated, which intensifies concern and implies that outcomes will matter beyond the two parties. These rhetorical choices raise the emotional impact while focusing the reader on legal, ethical, and practical stakes, guiding attention toward sympathy for rights holders, scrutiny of the company’s practices, and interest in the legal outcome.