Loading...
OpenAI using Oxford University's library to train AI models
Internal documents from Oxford reveal concerns among staff

OpenAI using Oxford University's library to train AI models

Sep 26, 2026
05:01 pm

What's the story

The University of Oxford has allowed OpenAI, the company behind ChatGPT, to use historical texts from its Bodleian library for training its artificial intelligence (AI) models. The move comes as tech companies are increasingly looking at academic institutions for new data sources. According to internal documents, the digitized Bodleian material has been used to "populate the OpenAI training set."

Collaboration details

Oxford's partnership with OpenAI

In March 2025, Oxford announced a partnership with OpenAI to digitize texts from its famous library.

The university said this would make the content more accessible to students and researchers.

However, the initial announcement didn't mention that this material would be used for training OpenAI's models, which learn by being fed large amounts of data.

AI ethics

OpenAI's commitment to diversity and inclusivity in AI training data

An OpenAI spokesperson expressed pride in ensuring "the AI models of today preserve the world's historical knowledge for the future."

They stressed that with over a billion people using their technology in everyday life, it is important to reflect different cultures, histories, and perspectives.

This statement highlights OpenAI's commitment to maintaining diversity and inclusivity in its AI training data.

ADVERTISEMENT

Backlash

Concerns among Oxford staff about reputational risks

Internal documents from Oxford reveal concerns among staff, including Bodleian governance committee members, about the reputational risk of partnering with OpenAI.

They also raised questions about how a deal involving an energy-intensive technology could affect the university's environmental commitments.

These concerns highlight the potential downsides of such collaborations in terms of public perception and sustainability goals.

ADVERTISEMENT

Data acquisition

The trend of AI developers seeking data from physical collections

As AI-generated material saturates scraped websites, developers are turning to physical, often historical, book collections for fresh data.

OpenAI has made similar deals with US research libraries under a project called NextGenAI. Oxford is the only UK member of this project.

By June 2025, 125,000 images scanned from historical dissertations had been shared with OpenAI from the Bodleian collection.

Digitization plans

Plans for public access to digitized texts

The OpenAI contract with Oxford opens up the possibility of mass digitization for Bodleian's 23 million-item collection.

A university spokesperson said the amount of text being digitized is "modest in scale" and only includes out-of-copyright material.

They also clarified that Bodleian retains rights to scans and will begin publishing them openly online within months, ensuring public access to these historical texts.

ADVERTISEMENT