Skip to main content

GLAM-E Lab Response to IPO, DSIT and DCMS Copyright & AI Consultation, February 2025

Dr. Andrea Wallace

Table of Contents

About us

The GLAM-E Lab is a joint initiative between the University of Exeter and New York University School of Law to improve the legal, technical and ethical landscape for open access to cultural heritage. We work directly with galleries, libraries, archives, museums, community organisations, data aggregators and other heritage platforms to develop open access programmes, capacity and expertise within the cultural sector. The solutions created are then integrated into model policies, guidance and tools that can be adopted by others. All GLAM-E Lab materials are published on the website as CC BY or CC0.

In addition to commissioned work by sector leaders like the National Lottery Heritage Fund, Europeana and Flickr Foundation, we have worked with cultural institutions and organisations around the world on the copyright, digitisation and open access challenges they face. We recently released an Open GLAM Toolkit, in partnership with the Royal Albert Memorial Museum & Art Gallery and Wikimedia UK, informed by these collaborations.

Overview

We welcome the opportunity to respond to the consultation. Much is at stake for the UK’s creative industries and cultural heritage sector, including significant gains. Indeed, there are vast amounts of out-of-copyright materials housed within cultural institutions which can be digitised for text and data mining, new research, creativity and economic growth. We see a crucial opportunity to encourage research and innovation, support the digital transformation of the cultural sector, protect the livelihoods of living creators and our vibrant creative industries, and produce high-quality data for AI development.

Our response focuses on the potential of these public domain materials for the development of world-leading AI models and new creative frontiers. For additional information or discussion of this submission, please contact Dr Andrea Wallace, UK Director of the GLAM-E Lab: a.wallace@exeter.ac.uk

We recommend a strategy that enables AI model training on out-of-copyright materials, drives resources and new investment to the cultural heritage sector, and enables human-centred creativity and our creative industries to flourish:

  • Clarify that no new copyright subsists in faithful reproductions of public domain works or other non-original materials, such as data and metadata.

  • Require the publication or deposit of such materials in high-quality without unnecessary restrictions on reuse (*e.g.*, contractual or otherwise).

  • Support AI partnerships with cultural heritage institutions and organisations to build capacity and expertise for everyone involved.

  • Install transparency measures that ensure attribution and sourcing and mitigate bias concerns to promote greater trust and transparency.

  • Renew and expand funding to the cultural heritage sector to enable them to fulfil public missions in the digital era.

Support

We note the consultation makes no mention of the public domain. This demonstrates thinking that the public domain is not relevant to the question of copyright and AI, which is a barrier to the consultation’s goals.

At present, it is common practice to assert new copyrights in the non-original reproductions of public domain works despite their failure to meet an essential requirement for protection: originality. Such copyright claims are made by galleries, libraries, archives, museums, photographers, commercial picture libraries, digitisation partners, and others involved in the reproduction of public domain collections.

This practice undermines the full economic and social potential of the public domain and prevents it from serving the purpose intended by copyright law. It exacerbates the imbalance of lawful access for the public, as well as for AI training and development on creative content, and significantly limits investment, innovation and growth in the UK.

In this spirit, we agree with the consultation objectives on clarity:

The copyright framework should provide legal clarity on what is and is not protected by copyright, for the benefit of right holders and users.

The Government should:

  1. reiterate that no new copyrights or related rights arise in non-original materials resulting from the reproduction of public domain works (e.g., images, models, text, data, metadata); and

  2. require such materials to be released in high-quality to the public for free and unrestricted reuse.

As the consultation highlights, (1) is already the case under UK copyright law. Yet most cultural institutions claim the law is uncertain, assert new copyrights, publish low-resolution images online and charge costly fees for use requests. Some accept that no copyright arises yet disable downloads on the website and rely on contract law to replicate copyright-based licensing services. This widespread lack of compliance with copyright law is a significant problem.

Acknowledging this, Labour’s Plan for the Arts, Culture and Creative Industries commits to developing an open data policy to ‘ensure that national museums and galleries allow everyone to download high-resolution images of works of art that are in the public domain, free of charge for non-commercial use’. However, this limitation on non-commercial use fails open data standards and raises barriers to bringing the full economic potential of the public domain, including for AI development. Open data means data can be freely used, re-used and shared by anyone for any purpose, including commercial use.

To date, the Government approach has been to leave decisions on copyright and licensing with institutions as an operational matter. But institutions do not have operational discretion to ignore UK copyright law. Institutional desires for licensing revenue cannot remove works from the public domain.

In addition to contrevening copyright law, institutional attempts to impose licensing obligations on public domain works are counterproductive. Studies consistently make two key findings about institutional services that license the public domain. First, image licensing services do not return profit. Some bring in nominal income, but most operate at a loss. Revenue is also typically generated by a small set of popular images, rather than distributed across the entire collection. Such fees are largely paid by researchers, often out-of-pocket.

Second, the real potential for income comes through the brand licensing and new partnerships made possible through open licensing. More than 1,700 cultural institutions, including many in the UK, have released more than 100 million digitised public domain collections for free and unrestricted reuse. These cultural institutions have continued to monetise their public domain assets—the difference is that now everyone else can, too.

This issue must be resolved. We need greater clarity and legal certainty to protect the public domain from being re-privatised, including when cultural heritage collections are digitised or used for AI. Removing any ambiguity will send a clear message on what is required to comply with UK copyright law while opening more high-quality data for AI training and bolstering respect for living creators and in-copyright works.

A clear statement supports the consultation objectives on interoperability:

We want to ensure that the UK’s copyright provisions are internationally interoperable and do not lead to unreasonable burdens for AI providers, which often operate across multiple jurisdictions.

Without clarity, the UK will fall further behind other countries that are leading the way. US federal courts have clarified that such reproductions are not sufficiently original for new copyright protections.The EU is even further ahead with policy and legislation that:

  • Prohibit new assertions of copyrights or related rights in such assets

  • Expand obligations to release them for public reuse

  • Build a common data space where researchers, SMEs and AI companies can find and freely use them

  • Make significant long-term investments in the digital transformation of the cultural sector

  • And, as a result, produce high-quality data for various sectors, including AI.

These initiatives recognise the importance of digital heritage assets and their reuse to invigorating the cultural economy and creative industries. They pave the way for the monumental and unprecedented impact of reusing cultural assets for innovative and creative services and products to improve investment, stability and resilience in various sectors.

Given the immense value of the public domain, allowing institutions to ignore the balance inherent in copyright law results in untold public welfare loss for the UK—and not just economically, but socially. It even limits the Government’s own return-on-investment given that significant funds have been invested in the digital transformation of the cultural sector in programmes like Towards A National Collection and the Museums Data Service without resolving the issue.

To be effective, there must be obligations to release high-quality data without new unnecessary restrictions on public reuse. New projects and partnerships that digitise collections cannot be allowed to lock them back up for their own private use using contractual terms or other restrictions.

Indeed, under the consultation’s preferred B4 Option 3, cultural institutions will continue to make improper copyright assertions and simply opt-out of AI training. Copyright issues and longstanding inequities will continue to pervade digital heritage and access to public domain collections at great cost to UK economies, industries and knowledge systems.

This strategy can help achieve the consultation’s goals:

A successful approach will be one that encourages more AI model training to take place in the UK and enables rights under UK copyright law to be respected. This ensures protections for the creative industries alongside an internationally competitive and interoperable regime for AI training.

An approach that protects and liberates the public domain will produce an immense volume of high-quality data for wider public reuse and encourage more AI model training to take place in the UK. Mutually beneficial partnerships among the cultural heritage sector, creative industries and AI companies are possible and can flourish. Participatory approaches to digitisation and AI development will enable the exchange of skills, education and literacy across all partners and secure cultural institutions in their role as trusted data intermediaries for cultural content.

Yet, the Government cannot leave the digital transformation that we desperately need to AI companies alone. We welcome the recent announcement of funding allocated to arts and culture. Further investment must be dedicated to the sector’s digital transformation. After 14 years of austerity, our cultural sector has reached a critical point. Museums, archives and libraries are shutting down. The UK’s world-leading universities are gutting Arts and Humanities degrees and departments. And yet this is the very expertise so crucial to the trajectory of AI. The cultural sector brings the humanistic perspectives and context necessary to ground AI research and innovation and support ethical AI. They must be adequately funded and have staff in the room to bring this expertise to the issues at play, including those related to bias, inaccuracies and harm found in cultural heritage data and collections.

We stand beside the government in their effort and propose an approach that operationalises the UK’s out-of-copyright collections to achieve its goals. We have a chance to support a thriving economy, attract investment in the UK, catalyse the digital transformation of the cultural sector, and unlock the vast potential of these cultural assets — all while protecting the livelihoods of living creators and their ability to be remunerated as copyright law intended.

  • 1

    An example of a faithful reproduction includes a faithful photographic reproduction of an existing painting, print, drawing or sculpture.

1

‘To be so protected it must be original, in the sense that it is the intellectual creation of its author, reflecting their personality and creative choices.’ p 23. See also UK Intellectual Property Office, ‘Copyright Notice: digital images, photographs and the internet’; THJ Systems Ltd v Sheridan [2023] EWCA Civ 1354.

1

Creating Growth: Labour’s Plan for the Arts, Culture and Creative Industries, page 33.

1

See Kathryn M. Rudy, ‘The true costs of research and publishing’ (Times Higher Education 2019).

1

Douglas McCarthy and Andrea Wallace, Open GLAM Survey, https://survey.glamelab.org/.

1

US courts have held photographic reproductions, scans and models of both 2D and 3D works are non-original. See Bridgeman Art Library, Ltd v Corel Corporation, 25 F.Supp. 2d 421 (S.D.N.Y. 1998); Bridgeman Art Library, Ltd v Corel Corporation, 36 F. Supp. 2d 191 (S.D.N.Y. 1998); Meshworks, Inc v Toyota Motor Sales USA, Inc, 528 F.3d 1258 (10th Cir. 2008); President & Fellows of Harvard College v Ellmore, No. 15-CV-00472-RB-KK, 2016 WL 7494272 (D.N.M. May 19, 2016); President & Fellows of Harvard College v Ellmore, 222 F. Supp. 3d 1050 (D.N.M. 2016); US Copyright Office, Compendium of US Copyright Practices § 909.3 (3d ed. 2014) (instructing the Copyright Office not to register works “if it is clear that the photographer merely used the camera to copy the source work without adding any creative expression to the photo”); Policy Decision on Copyrightability of Digitized Typefaces, 53 Fed. Reg. 38110, 38113 (Sept. 29, 1988) (stating that digitization fails to create authorship and rather “digitized version is a copy of the pre-existing work and would be protected as such, but no new work of authorship is created[…]. Protection depends on the status of [the pre-existing work]; digitization does not add any new authorship”).

  • 1

    Article 14 of the Directive 2019/790 on Copyright in the Digital Single Market prohibits any non-original media resulting from the reproduction of out-of-copyright artworks from being subjected to new copyrights or related rights. See also Case C-469/17, Funke Medien NRW GmbH v Bundesrepublik Deutschland (29/7/2019) ECLI:EU:C:2019:623, paras 19-23.

  • 1

    Directive 2019/1024 on Open Data and the Re-use of Public Sector Information OJ L 172/56.

  • 1

    Commission Recommendation (EU) 2021/1970 of 10 November 2021 on a common European