Encyclopaedia Britannica and Merriam-Webster have filed a lawsuit against OpenAI, accusing the firm of infringing on the copyright of nearly 100,000 articles by using them to train large language models without proper authorization.
The case marks a significant escalation in the growing legal battle between content creators and artificial intelligence companies, as publishers seek to define how intellectual property laws apply in the age of generative AI.
According to the complaint, the publishers allege that OpenAI incorporated substantial portions of their copyrighted material into its training datasets, enabling its models to generate responses that closely resemble or draw from their original content. They argue that this constitutes unauthorized use and undermines the value of their intellectual property.
The lawsuit reflects broader concerns among traditional knowledge institutions, which have invested decades in building curated, authoritative databases. Encyclopaedia Britannica, one of the oldest reference publishers in the world, and Merriam-Webster, a leading dictionary publisher, both maintain extensive archives of educational and linguistic content that are widely used across academic and professional settings.
At the centre of the dispute is the question of whether using copyrighted material to train AI systems falls under fair use or requires licensing agreements. OpenAI and other AI developers have generally argued that training models on large datasets, including publicly available text, is a transformative process that does not directly reproduce original works. Publishers, however, contend that the scale and nature of this use go beyond acceptable limits.
The case is part of a wave of legal actions targeting AI companies over data usage. Similar lawsuits have been filed by authors, news organisations and media companies, all seeking clarity on how their content can be used in training algorithms. These cases could set important legal precedents that shape the future of the AI industry.
Legal experts say the outcome could have far reaching implications. If courts rule in favour of publishers, AI companies may be required to secure licenses for training data or compensate content owners, potentially increasing operational costs and slowing development. On the other hand, a ruling that supports AI firms could reinforce the current model of using large scale datasets, accelerating innovation but raising ongoing concerns about intellectual property rights.
The lawsuit also highlights the tension between technological advancement and the protection of creative and educational work. As AI systems become more capable, the demand for high quality training data has increased, bringing them into direct conflict with organisations that produce and own such content.
OpenAI has not publicly detailed its legal response to the claims, but the company has previously indicated a willingness to work with publishers through partnerships and licensing agreements. In recent months, several AI firms have entered into deals with media organisations to access content legally, suggesting a possible path toward collaboration rather than conflict.
For Encyclopaedia Britannica and Merriam-Webster, the case represents an effort to assert control over how their content is used in a rapidly evolving digital landscape. For the AI industry, it underscores the urgent need for clearer regulatory frameworks governing data usage and intellectual property.
As the case progresses, it is likely to become a landmark moment in defining the relationship between AI development and content ownership, with implications that extend across technology, media and education sectors worldwide.