Encyclopedia Britannica Sues OpenAI: Nearly 100,000 Articles Used for Training, Illusionary Outputs Also Misappropriated Its Name

According to 1M AI News, last Friday, Encyclopedia Britannica and its subsidiary Merriam-Webster filed a lawsuit against OpenAI in the Manhattan Federal Court, accusing the latter of unauthorized use of its online articles, encyclopedia entries, and dictionary definitions to train ChatGPT. The complaint alleges that OpenAI illegally copied nearly 100,000 Encyclopedia Britannica articles to train the GPT language model, and ChatGPT is capable of generating responses almost identical to encyclopedia entries and dictionary definitions, diverting users who would have otherwise visited its website.

Encyclopedia Britannica also accuses OpenAI of infringing its trademark rights: first, by implying authorized use of its content, and second, by incorrectly citing Encyclopedia Britannica as a source within AI-generated false "illusions." Encyclopedia Britannica is seeking compensation (amount unspecified) and a restraining order against further infringement. An OpenAI spokesperson responded, stating, "Our model drives innovation, is trained on publicly available data, and is built on fair use principles." Previously, Encyclopedia Britannica filed a similar lawsuit last year against AI search startup Perplexity AI, which is still pending.