Oxford's bodleian library shares historical texts with OpenAI for AI training

OpenAI has been using digitized historical texts from Oxford's Bodleian Library to train its AI models, according to a report by The Guardian. The collaboration, announced in March 2025, involves the digitization of 125,000 pages of 19th- and 20th-century doctoral theses and other rare materials. A spokesperson for OpenAI stated that the partnership helps preserve cultural heritage for global users. The Bodleian Library, the second-largest in the UK, has also shared a collection of 10,000 16th-century broadside ballads. Concerns were raised by faculty members about reputational risks and environmental impact. The library maintains control over the digital resources and plans to make them publicly available soon. The project is part of OpenAI's NextGenAI initiative, which includes partnerships with U.S. institutions like MIT and Caltech.