Leiter Reports: A Philosophy Blog

News and views about philosophy, the academic profession, academic freedom, intellectual culture, and other topics. The world’s most popular philosophy blog, since 2003.

  1. Giovanni Molteni Tagliabue's avatar
  2. Fabien Muller's avatar
  5. Saul Smilansky's avatar
  6. Dan Dennis's avatar

    Some background: https://www.theguardian.com/education/2026/may/12/thousands-of-university-of-nottingham-staff-told-they-are-at-risk-of-redundancy Not only does Nottingham University have a good academic reputation, the city of Nottingham has a great…

  7. Jacob Barrett's avatar

Cambridge University Press now asking authors whether they want to license their publications for LLMs

Brian Leiter

Curious to hear what readers think about this development.  A law professor elsewhere (who published a book with CUP a few years back) shared the email he received from the press:

Following developments in artificial intelligence (AI) during 2022 and 2023, Cambridge University Press has begun to receive content licensing requests from providers of generative artificial intelligence. We value our authors and editors and want to collaborate with you to establish a responsible and transparent way forward in this evolving space. As such, we are contacting you to provide more information on our approach and to request your permission for your work to be included in new licensing routes.

You will shortly receive an email link to a digital addendum to your author or editor contract with us for [book name removed] from a system called Ironclad / HelloSign. If you agree that your work can be included in generative AI licensing deals, click the link in this email to sign the addendum. Please note that this email may be in your junk or spam folder.

Context

Generative AI technologies offer opportunities and risks. The use of high-quality scholarly content in training generative AI models may improve the quality and accuracy of tools that are increasingly going to be used in everyday life. There may also be opportunities for your content to have greater visibility and impact if it is properly cited and attributed by AI tools, and to drive the creation of new and innovative products for a range of audiences. Many of our existing relationships with third-party vendors and discovery services, which help to distribute and sell publications, are likely to involve AI services in that process in future, too.

As use of AI grows, we believe that use of the content we publish should be:

  • Subject to clear principles around attribution
  • Governed by formal licensing arrangements with generative AI providers
  • Founded on obtaining the relevant permission from all rightsholders
  • A source of fair remuneration for author and publisher

It is a priority for us to work towards these guidelines as we negotiate and agree licences with prospective licensing partners.

What this means for you

Use of your content

If your work is part of a generative AI licensing agreement, it could be used for:

  • Training and testing the foundational models that are then used to create, for example, personal assistant and chatbot tools or discoverability summaries
  • As part of banks of authoritative content that are used, on a perpetual basis, to check and verify the accuracy of information provided by AI tools

Other use cases will emerge as technology evolves, and our aim is to clearly define and set limits on intended uses of content published by Cambridge University Press whenever we set up new licensing relationships.

To set expectations with prospective partners, and to ensure the responsible use of the content we publish, we are currently focusing on the following principles in our negotiations with prospective licensing partners:

  • Limits on the amount of text that can be reproduced
  • Requirements that work is appropriately cited
  • Removal of third-party content such as figures, images and illustrations
  • Limits on the ability to adapt or modify your work, or to create new works based on it
  • Limits on sub-licensing of the work, and requirements that it is kept confidential and secure
  • Removal of content once a licensing term has ended

For more information on how your work may be used, you can see our FAQs here.

Financial remuneration

Given the scale of the new opportunities these licensing deals present, we want to ensure rightsholders are fairly remunerated. As such, if your work is used in a generative AI licensing deal, we will be paying a royalty of 20%. For more information on how royalties will be calculated, please see our Author FAQs.

How to send us your response

You will shortly receive an email link to a digital addendum to your author or editor contract with us from a system called Ironclad / HelloSign. If you agree that your work can be included in generative AI licensing deals, click the link in this email to sign the addendum. Please note that this email may be in your junk or spam folder.

Please note that all rightsholders in each work will need to provide a digital signature before it can be included in an AI licensing deal, and if you have multiple books with us, you will receive an addendum for each one. Your timely response in this matter will ensure your content can be included.

What will happen if I don’t sign the addendum?

If you are an author or an editor and don't sign the relevant author or editor addendum, your work will not be included in generative AI licensing deals, and, depending on how technology and use cases develop, this may also limit our ability to license it to existing licensing partners or vendors who go on to use AI in the future.

Feedback and queries

The generative AI space is changing rapidly, and we will be providing updated information as and when it becomes available. Please check our FAQs for more information and the latest updates.

If you would like to contact us with questions or to discuss the addendum, please email authorrights@cambridge.org. This will be the fastest way to ensure that your queries are passed on to the relevant teams.

If you believe you are receiving this email and addendum in error, please contact us at authorrights@cambridge.org.

Many thanks,

Cambridge University Press

Leave a Reply

Your email address will not be published. Required fields are marked *

3 responses to “Cambridge University Press now asking authors whether they want to license their publications for LLMs”

  1. Publishing Veteran

    Publishers are trying to learn from their mistakes. They were slow to see the financial threats to their existence from the Web. Though not nearly as badly damaged as the newspaper business, the idea that “information wants to be free” was quickly embraced in academia, and that has placed journal subscription models under pressure for most of the past 25 yrs. The uptake of Open Access has compounded the challenge for subscription revenues, but the way the big players have harnessed the APC mechanism has allowed many of them to grow overall revenues despite negative trends for subscriptions.

    What is happening now is that the bigger publishers are trying to protect themselves from having their content taken over by companies like OpenAI without any licensing fees coming into play. The Open Access Creative Commons licenses, could be interpreted as alllowing wholesale use of OA papers for LLM purposes, so moves of the kind that CUP is making are attempts to foreclose that outcome. Informa, the owner of Routledge and Taylor & Francis, just did a deal for $10 mil with Microsoft to supply content for ChatGPT. In the world of news, NewsCorp is doing a 5-yr deal with OpenAI that could reach a value of $250 mil.

    Reply
  2. Dean C. Rowan

    I am agnostic about whether or not copyrighted work should be available to LLMs. On the one hand, I'd say that it's a no brainer: of course they should be. LLMs entail substantial transformations of the copyrighted material, and the encroachment on the market for published scholarship is nil. But it's entirely possible that I'm missing something.

    My real concern is with the corpora that will comprise these LLMs. The value of "big data" is in its capacity to represent fairly the random gamut of content. (The Google Books project, however successful it was and has been, failed miserably in this regard, because it scanned large research university collections.) We already know that corpora embody in some ways the biases inherent in their texts. And now CUP is proposing a purely selective opt-in approach that is sure to skew the representation of expression.

    Meanwhile, the only mistakes commercial publishers are trying to correct are those that compromise their profits. I know that, PV, and you know that, too. Scholarship is of little to no interest to them.

    Reply
  3. Keith Douglas

    I would be also worried about the resulting models and the so-called hallucinations. From the perspective of my current profession, cyber security, these are data integrity losses of particularly egregious and (sometimes) dangerous nature. By agreeing to include the work in the training set, then you also risk that it will be used in ways that are nonsense, false, etc. This is independent of the bias concerns that was previously mentioned. There is no known way to prevent these and there is some research that suggest that they are unavoidable at a rather profound level. (Turing's bit about intelligence requiring fallibility may be in play!) Admittedly that the world is full of other falsities and nonsense, however I suspect -and would love for the cognitive psychologists to figure out – that the so-called "ELIZA Effect" makes it more likely that people will believe (for example) chatbots.

    —–
    KEYWORDS:
    Primary Blog

    Reply

Designed with WordPress