University of Oxford Partners With OpenAI to Digitize Texts
The University of Oxford is digitizing historical Bodleian Library texts to help OpenAI train its AI models while increasing public access to out-of-copyright materials.
The University of Oxford has partnered with OpenAI to digitize historical texts from the Bodleian Library for use in training artificial intelligence models. The collaboration, which began in March 2025, has already resulted in the transfer of 125,000 images of historical dissertations and 10,000 16th-century broadside ballads to the AI company.
Oxford is the sole United Kingdom member of OpenAI's NextGenAI project, joining United States institutions such as MIT and Caltech. The university maintains that the project aims to make out-of-copyright materials more accessible to students and researchers, and the scans will be published openly online in the coming months.
Despite the public goals, internal documents indicate that university staff have raised concerns regarding environmental impacts and potential reputational risks. The university has defended the scope of the project, stating the scale of digitization is modest and that the Bodleian retains the rights to the scans.