In March 2025, #Oxford announced that it was working with #OpenAI to digitize #PublicDomain #books from the #BodleianLibrary.
https://www.bodleian.ox.ac.uk/about/media/oxford-and-openai-launch
The announcement says that the digitized works "will become searchable and accessible for students and researchers worldwide." To me that sounds like #OpenAccess.
Q1: Can anyone confirm that Oxford means OA? I have to ask, since Oxford knows the term and apparently chose not to use it.
If the answer is yes, then OpenAI gets a corpus of important new texts to train on and the world gets OA. If so, then all of OpenAI's rivals get the same corpus to train on. If so, then the benefit to OpenAI worth the cost of the digitization is not exclusive access. It's just a head start.
Q2: Can anyone confirm that the chief benefit to OpenAI here is just a head start in training on these texts?
Yesterday (more than 1.5 years later) The Guardian reported on the project and doesn't answer my two questions above.
https://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt
Nor does it answer my Q3: Have any of these text already become OA, available to human readers, AI companies, and other entities beyond Oxford and OpenAI?
#AI #Digitization
petersuber
npub1py…n7td9
2026-09-27 16:54:56 UTC
Author Public Key
npub1py4nz04lczsv4fyr3lywsjss5eqedlckk2npue8wsy7fwzgzrrks6n7td9Published at
2026-09-27 16:54:56 UTCEvent JSON
{
"id": "b845e5aa6f73b836045e49a88fed06906b7e9e8cbcb12a040b7f805c425752cc",
"pubkey": "092b313ebfc0a0caa4838fc8e84a10a64196ff16b2a61e64ee813c97090218ed",
"created_at": 1790528096,
"kind": 1,
"tags": [
[
"t",
"openai"
],
[
"t",
"digitization"
],
[
"t",
"publicdomain"
],
[
"t",
"oxford"
],
[
"t",
"openaccess"
],
[
"proxy",
"https://fediscience.org/@petersuber/117344049339922612",
"web"
],
[
"t",
"ai"
],
[
"t",
"books"
],
[
"t",
"bodleianlibrary"
],
[
"proxy",
"https://fediscience.org/users/petersuber/statuses/117344049339922612",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://fediscience.org/users/petersuber/statuses/117344049339922612",
"pink.momostr"
],
[
"-"
]
],
"content": "In March 2025, #Oxford announced that it was working with #OpenAI to digitize #PublicDomain #books from the #BodleianLibrary. \nhttps://www.bodleian.ox.ac.uk/about/media/oxford-and-openai-launch\n\nThe announcement says that the digitized works \"will become searchable and accessible for students and researchers worldwide.\" To me that sounds like #OpenAccess. \n\nQ1: Can anyone confirm that Oxford means OA? I have to ask, since Oxford knows the term and apparently chose not to use it.\n\nIf the answer is yes, then OpenAI gets a corpus of important new texts to train on and the world gets OA. If so, then all of OpenAI's rivals get the same corpus to train on. If so, then the benefit to OpenAI worth the cost of the digitization is not exclusive access. It's just a head start. \n\nQ2: Can anyone confirm that the chief benefit to OpenAI here is just a head start in training on these texts?\n\nYesterday (more than 1.5 years later) The Guardian reported on the project and doesn't answer my two questions above.\nhttps://www.theguardian.com/technology/2026/sep/26/oxford-university-bodleian-library-open-ai-chat-gpt\n\nNor does it answer my Q3: Have any of these text already become OA, available to human readers, AI companies, and other entities beyond Oxford and OpenAI?\n\n#AI #Digitization",
"sig": "e28b21acc25dfc08df94ab6ef949a452fbb42693658a4db03834483ac0fa843e24ca90104468cc6ce1af1e50bb27d97c8bab54d9bf2eea0a2f7a43eae16963c5"
}