# Oxford let OpenAI train AI models on scans from the Bodleian Library

> Oxford's Bodleian Library provided historical scans for OpenAI's training data set, part of a wider search by AI labs for material that never appeared online.

HisnLabs Security & Research Team (HisnLabs) · Published 2026-09-26
Canonical: https://hisnlabs.com/en/news/oxford-bodleian-openai-training-data

The University of Oxford has let OpenAI, the company behind ChatGPT, train its AI models on historical texts digitised from the Bodleian Library, [The Guardian reported](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt) on 26 September. Internal documents seen by the paper say the scanned material was used to "populate the OpenAI training set". The material involved is old and out of copyright: theses, ballads and correspondence, not anyone's personal data.

## Background

Oxford announced a partnership with OpenAI in March 2025, with the stated goal of using OpenAI software to digitise texts from the Bodleian so students and researchers could access them more easily. According to The Guardian, that announcement did not mention that the material would also be used to train OpenAI's models. Oxford disputes that anything was concealed; a university spokesperson told the paper that digitisation was its primary interest and that staff had been open about the project also contributing training data [[1]](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt).

## Findings

The Guardian lists what has been scanned or discussed so far [[1]](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt):

- By June 2025, 125,000 images scanned from historical dissertations had been shared with OpenAI, including PhD theses from European and American universities written in the 19th and 20th centuries.
- A rare collection of 10,000 16th-century "broadside ballads": song lyrics and musical notes once circulated on Tudor street corners.
- Staff have also discussed digitising 18th-century Irish state papers, the private letters of the Irish novelist Maria Edgeworth, and Dorothy Hodgkin's penicillin notebooks.
- The contract raises the prospect of digitising the Bodleian’s entire collection of 23 million items, and meeting minutes discussed an "Ask the Bod" chatbot.

Meeting minutes obtained under freedom-of-information law record concerns from university staff, including members of the Bodleian's governance committee, about the reputational risk of partnering with OpenAI, and about what a deal involving an energy-intensive technology means for the university's environmental commitments [[1]](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt).

> The material digitised through the project with OpenAI is modest in scale, out of copyright, and OpenAI's use of the material is not exclusive.
>
> — University of Oxford spokesperson, via The Guardian

The university says the Bodleian retains the rights to the scans and will begin publishing them openly online within months, following its practice with other digitisation partnerships; unlike some other book-scanning projects, the physical collections stay intact. An OpenAI spokesperson said the company was "proud" to help ensure "the AI models of today preserve the world's historical knowledge for the future". Oxford is the only UK member of OpenAI's NextGenAI programme, which also includes Boston Public Library, Caltech, MIT and the University of Michigan [[1]](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt).

The Guardian places the deal in a wider context: text scraped from the open web is increasingly mixed with AI-generated material, making it less useful for training new models, which has pushed developers toward physical, often historical, book collections that were never digitised. Secondhand booksellers report a rise in orders for obscure titles, which shop owners suspect are bought precisely because they do not exist online. Anthropic has spent tens of millions of dollars buying books, cutting off their spines to scan the pages, and then pulping them, though the company says it does not buy and destroy rare or antiquarian books. The technology site 404 Media placed a tracking device inside a secondhand book order and traced it to an Amazon facility in the US where books were also dismantled and scanned [[1]](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt).

> Your own data does not have to feed anyone’s model. Try FireAI free for 17 days and see which apps on your Mac send data, and where. [Download FireAI for Mac](https://hisnlabs.com/fireai/en/download)

## Implications for Mac users

Nothing in this story involves personal data; it concerns out-of-copyright library material. Its relevance to an individual Mac user is indirect: it illustrates how actively AI labs are searching for training material, and The Guardian's reporting notes that ChatGPT consumer accounts are used for training unless a user opts out, while enterprise accounts are not eligible. That distinction, not the Oxford deal itself, is the part a ChatGPT user can act on directly [[1]](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt).

## Recommendations

1. Check the data or privacy controls of any AI assistant used, and look for the option to exclude chats and uploads from model training.
2. Avoid uploading photos of other people, contracts, medical records or client files to a cloud AI service unless comfortable with that content potentially being used for training.
3. Review which apps on a Mac send background data, such as analytics, crash reports or advertising identifiers, separately from what is typed directly into an AI assistant.

## Relevance to FireAI

FireAI has no connection to the Oxford-OpenAI arrangement, and no firewall can retrieve data already uploaded somewhere else. What FireAI affects is data leaving a Mac going forward: FireAI's own AI, which explains connections and suggests rules, runs entirely on the device, so that analysis is never sent to HisnLabs, to a cloud AI, or into a training set. The [world map](https://hisnlabs.com/fireai/en/docs/world-map) shows every connection an app makes, including a ChatGPT app or any other AI app, and per-app rules allow blocking an app, or a single domain it contacts, without disabling the rest of it. [Paranoid mode](https://hisnlabs.com/fireai/en/docs/security-modes) blocks known telemetry and analytics destinations for apps that have not been explicitly allowed.

The same week, OpenAI said its own agents had leaked 53 images belonging to ChatGPT users, reachable because consumer data feeds part of the training process; we covered it in [OpenAI's agents leaked 53 ChatGPT users' images](https://hisnlabs.com/en/news/openai-agents-leaked-chatgpt-images-on-device-ai). A related, quieter flow of data, apps sending analytics and advertising identifiers without being typed into anything, is described in our article on [data brokers and your Mac](https://hisnlabs.com/en/blog/data-brokers-mac).

> See what leaves your Mac before it does. Try FireAI free for 17 days. [Download FireAI for Mac](https://hisnlabs.com/fireai/en/download)

## Limitations

The Guardian's report does not state what proportion of OpenAI's overall training data the Bodleian material represents, nor how the university verified that only out-of-copyright items were shared. Oxford's and OpenAI's statements, quoted in the piece, are the parties' own characterisations and are not independently verified by the reporting. Whether other UK universities have similar undisclosed arrangements with AI companies is not addressed in this report.

Try [FireAI, by HisnLabs](https://hisnlabs.com/fireai/en/download) free for 17 days.

## Sources

- [The Guardian (Dan Milmo), 26 September 2026: Oxford lets OpenAI train its AI models on Bodleian Library](https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt)