The University of Oxford has established a partnership with OpenAI that includes the use of digitized historical texts from the Bodleian Library for training the company’s artificial intelligence models. The arrangement was initially announced in March 2025 as a digitization initiative designed to improve accessibility of the library’s materials for students and researchers, though the machine-learning component was not prominently highlighted in the original announcement.
By June 2025, approximately 125,000 images scanned from historical dissertations had been shared with OpenAI, including doctoral theses from European and American universities spanning the 19th and 20th centuries. The digitized materials also encompass a collection of 10,000 16th-century broadside ballads and other historical documents. Internal university meeting minutes, revealed through a freedom of information request, documented staff concerns regarding reputational implications and environmental impacts of the partnership.
Oxford is the sole United Kingdom participant in OpenAI’s broader NextGenAI project, which includes similar arrangements with American research institutions including Boston Public Library, Caltech, MIT, and the University of Michigan. Tech companies have increasingly turned to physical book collections as training data sources because internet-scraped content is becoming saturated with artificial intelligence-generated material, reducing its utility for model development.
A university spokesperson characterized the digitization effort as modest in scale and limited to out-of-copyright materials. The Bodleian Library retains ownership of the digital scans and plans to publish them openly online within months. The agreement also preserves the physical collection intact, distinguishing it from other arrangements where rival AI companies have purchased and destroyed books after scanning their contents. University officials emphasized that the digitization project will ultimately enhance public access to rare historical materials.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI