The Obligations of Providers of General-Purpose AI Models

The Obligations of Providers of General-Purpose AI Models external link

Veale, M. & Quintais, J.

Fortcoming in: Zanfir Fortuna, Malgieri, González Fuster and Mantelero (eds.) The Artificial Intelligence Act — A Thematic Commentary, Hart Publishing, 2026. This version: November , 2025

Abstract

During the legislative process, the EU Artificial Intelligence (AI) Act was amended to include provisions related to general-purpose AI (GPAI) models. These broadly relate to transparency towards downstream users and relevant regulators, in addition to obligations connected to intellectual property. In this paper, we provide detailed analysis of these new provisions in the context of current technological applications and emerging trajectories, connecting them to computing literature and practice, and the broader context of connected and adjacent legal regimes, in particular copyright and relevant emerging case law. We find that there are a significant number of inclarities, tensions and contradictions both within the text, between the text and other legal regimes, and between the text and guideline documents, such as the Code of Practice on General-Purpose AI and recent guidelines by the European Commission. We identify a range of issues with the scoping of the provisions which may undermine its policy goals and create loopholes for regulatory avoidance, such as those relating to non-commercial models, open-source models, and model finetuning along the value chain. We find that the Code of Practice contains significant omissions and misstatements, some of which may present a compliance risk for an entity choosing to rely on the Code. We do not consider the provisions on GPAI models which present a systemic risk, which are dealt with elsewhere in the volume which this work will form a part of.

Links

AI Act, code of practice, Copyright, Transparency

RIS

TY  - CHAP
AU  - Veale, M.
AU  - Quintais, J.
PY  - 2025
DA  - 2025-11-17
T1  - The Obligations of Providers of General-Purpose AI Models
UR  - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5744602
DO  - https://doi.org/10.2139/ssrn.5744602
AB  - During the legislative process, the EU Artificial Intelligence (AI) Act was amended to include provisions related to general-purpose AI (GPAI) models. These broadly relate to transparency towards downstream users and relevant regulators, in addition to obligations connected to intellectual property. In this paper, we provide detailed analysis of these new provisions in the context of current technological applications and emerging trajectories, connecting them to computing literature and practice, and the broader context of connected and adjacent legal regimes, in particular copyright and relevant emerging case law. We find that there are a significant number of inclarities, tensions and contradictions both within the text, between the text and other legal regimes, and between the text and guideline documents, such as the Code of Practice on General-Purpose AI and recent guidelines by the European Commission. We identify a range of issues with the scoping of the provisions which may undermine its policy goals and create loopholes for regulatory avoidance, such as those relating to non-commercial models, open-source models, and model finetuning along the value chain. We find that the Code of Practice contains significant omissions and misstatements, some of which may present a compliance risk for an entity choosing to rely on the Code. We do not consider the provisions on GPAI models which present a systemic risk, which are dealt with elsewhere in the volume which this work will form a part of.
K1  - AI Act
K1  - code of practice
K1  - Copyright
K1  - Transparency
ER  -

Save .RIS

Bibtex

Chapter{nokey,
	title = {The Obligations of Providers of General-Purpose AI Models},
	author = {Veale, M. and Quintais, J.},
	doi = {https://doi.org/10.2139/ssrn.5744602},
	year = {2025},
	date = {2025-11-17},
	abstract = {During the legislative process, the EU Artificial Intelligence (AI) Act was amended to include provisions related to general-purpose AI (GPAI) models. These broadly relate to transparency towards downstream users and relevant regulators, in addition to obligations connected to intellectual property. In this paper, we provide detailed analysis of these new provisions in the context of current technological applications and emerging trajectories, connecting them to computing literature and practice, and the broader context of connected and adjacent legal regimes, in particular copyright and relevant emerging case law. We find that there are a significant number of inclarities, tensions and contradictions both within the text, between the text and other legal regimes, and between the text and guideline documents, such as the Code of Practice on General-Purpose AI and recent guidelines by the European Commission. We identify a range of issues with the scoping of the provisions which may undermine its policy goals and create loopholes for regulatory avoidance, such as those relating to non-commercial models, open-source models, and model finetuning along the value chain. We find that the Code of Practice contains significant omissions and misstatements, some of which may present a compliance risk for an entity choosing to rely on the Code. We do not consider the provisions on GPAI models which present a systemic risk, which are dealt with elsewhere in the volume which this work will form a part of.},
	keywords = {AI Act, code of practice, Copyright, Transparency},
}

Save .bib

Voorbereid external link

van Daalen, O.

POM press, 2025, pp: 264, ISBN: 9789493434196

Abstract

Wetenschappers zijn het erover eens: wereldwijde catastrofe is nog nooit zo dichtbij geweest; de dreiging is nu groter dan tijdens de Koude Oorlog. Rechter-plaatsvervanger, advocaat, burgerrechtenactivist en journalist Ot van Daalen zag en voelde het overal om zich heen, en besloot het beest in de bek te kijken. Hij spitte overheidsdocumenten door, sprak unieke experts die normaal achter de schermen blijven, reisde het hele land door, bezocht bunkers en volgde survivaltraining. Zo kreeg hij een helder beeld van de rampscenario's voor Nederland - wat er kan gebeuren, hoe de overheid dan zal reageren en wat jij zelf kan doen om je beter voor te bereiden.

Links

https://voorbereidboek.nl/

RIS

Save .RIS

Bibtex

Save .bib

Tokenistic Decentralisation or Non-Tokenistic Distributism: Capitalist Blockchain Narratives and Varoufakis’s Alternative external link

Smethurst, R., Barbereau, T. & Bodó, B.

Triple C: Journal for a Global Sustainable Information Society, vol. 23, iss. : 2, pp: 320-337, 2025

Abstract

Yanis Varoufakis wrote a science fiction novel, Another Now: Dispatches from an Alternative Present, to encourage post-capitalist political projects in our so-called real world. Costa, a protagonist from the novel, invents a portal that enables communication between his universe and a parallel universe. The two universes diverged after the global financial crisis in 2008. Private money networks like Bitcoin emerged in Costa’s capitalist universe, while in the alternative universe, a post-capitalist society uses blockchain technology for “a plain vanilla public payments system”. Our essay draws a sophistic comparison between liberal-cum-libertarian blockchain narratives from our universe and the science-fictional blockchain narrative from Another Now. We distinguish tokenistic decentralisation (a liberal-cum-libertarian notion) and non-tokenistic distributism (a post-capitalist concept). Liberal-cum-libertarian narratives treat blockchain as a cause of decentralisation and self-sovereignty (individual empowerment). Varoufakis’s science-fictional narrative, by contrast, describes the use of blockchain for a distributist political cause.

Links

RIS

TY  - JOUR
AU  - Smethurst, R.
AU  - Barbereau, T.
AU  - Bodó, B.
PY  - 2025
DA  - 2025-11-10
T1  - Tokenistic Decentralisation or Non-Tokenistic Distributism: Capitalist Blockchain Narratives and Varoufakis’s Alternative
JO  - Triple C: Journal for a Global Sustainable Information Society
VL  - 23
IS  - 2
SP  - 320-337
UR  - https://triple-c.at/index.php/tripleC/article/view/1568
DO  - https://doi.org/10.31269/y8jpac38
AB  - Yanis Varoufakis wrote a science fiction novel, Another Now: Dispatches from an Alternative Present, to encourage post-capitalist political projects in our so-called real world. Costa, a protagonist from the novel, invents a portal that enables communication between his universe and a parallel universe. The two universes diverged after the global financial crisis in 2008. Private money networks like Bitcoin emerged in Costa’s capitalist universe, while in the alternative universe, a post-capitalist society uses blockchain technology for “a plain vanilla public payments system”. Our essay draws a sophistic comparison between liberal-cum-libertarian blockchain narratives from our universe and the science-fictional blockchain narrative from Another Now. We distinguish tokenistic decentralisation (a liberal-cum-libertarian notion) and non-tokenistic distributism (a post-capitalist concept). Liberal-cum-libertarian narratives treat blockchain as a cause of decentralisation and self-sovereignty (individual empowerment). Varoufakis’s science-fictional narrative, by contrast, describes the use of blockchain for a distributist political cause.
ER  -

Save .RIS

Bibtex

Article{nokey,
	title = {Tokenistic Decentralisation or Non-Tokenistic Distributism: Capitalist Blockchain Narratives and Varoufakis’s Alternative},
	author = {Smethurst, R. and Barbereau, T. and Bodó, B.},
	doi = {https://doi.org/10.31269/y8jpac38},
	year = {2025},
	date = {2025-11-10},
	journal = {Triple C: Journal for a Global Sustainable Information Society},
	volume = {23},
	issue = {2},
	pages = {320-337},
	abstract = {Yanis Varoufakis wrote a science fiction novel, Another Now: Dispatches from an Alternative Present, to encourage post-capitalist political projects in our so-called real world. Costa, a protagonist from the novel, invents a portal that enables communication between his universe and a parallel universe. The two universes diverged after the global financial crisis in 2008. Private money networks like Bitcoin emerged in Costa’s capitalist universe, while in the alternative universe, a post-capitalist society uses blockchain technology for “a plain vanilla public payments system”. Our essay draws a sophistic comparison between liberal-cum-libertarian blockchain narratives from our universe and the science-fictional blockchain narrative from Another Now. We distinguish tokenistic decentralisation (a liberal-cum-libertarian notion) and non-tokenistic distributism (a post-capitalist concept). Liberal-cum-libertarian narratives treat blockchain as a cause of decentralisation and self-sovereignty (individual empowerment). Varoufakis’s science-fictional narrative, by contrast, describes the use of blockchain for a distributist political cause.},
}

Save .bib

Postkantoor in Iceland: On a margin of appreciation for European IP Offices in assessing the descriptiveness of trademarks download

Westenbroek, T.F.

Berichten Industriële Eigendom, iss. : 5, pp: 194-201, 2025

Links

Westenbroek Postkantoor in Iceland BIE 2025_5

margin of appreciation, Trademark law

RIS

Save .RIS

Bibtex

Save .bib

Freedom of Political Expression and the Limits of Trademark Power: IKEA v. Vlaams Belang external link

Izyumenko, E.

Human Rights Here, 2025

Links

https://www.humanrightshere.com/post/freedom-of-political-expression-and-the-limits-of-trademark-power-ikea-v-vlaams-belang

Freedom of expression, Trademark law

RIS

Save .RIS

Bibtex

Save .bib

Opinie: De EMFA treedt in werking: kleine stappen in grote schoenen download

Bouchè, G.

Mediaforum, iss. : 5, pp: 169, 2025

Links

02_Opener_MF 2025-5

RIS

Save .RIS

Bibtex

Save .bib

Thuiskopieheffing is nog niet op haar retour external link

Valk, E.G.

IE-forum, 2025

Links

https://ie-forum.nl/artikelen/thuiskopieheffing-is-nog-niet-op-haar-retour

Copyright, thuiskopie

RIS

Save .RIS

Bibtex

Save .bib

Comparing the Right to an Explanation of Judicial AI by Function: Studies on the EU, Brazil, and China external link

Metikoš, L., Iglesias Keller, C., Qiao, C. & Helberger, N.

pp: 31, 2025

Abstract

Courts across the world are increasingly adopting AI to automate various tasks. But, the opacity of judicial AI systems can hinder the ability of litigants to contest vital pieces of evidence and legal observations. One proposed remedy for the inscrutability of judicial AI has been the right to an explanation. This paper provides an analysis of the scope and contents of a right to an explanation of judicial AI in the EU, Brazil, and China. We argue that such a right needs to take into account that judicial AI can perform widely different functions. We provide a classification of these functions, ranging from ancillary to impactful tasks. We subsequently compare, by function, how judicial AI would need to be explained under due process standards, Data Protection Law, and AI regulation in the EU, Brazil, and China. We find that due process standards provide a broad normative basis for a derived right to an explanation. But, these standards do not sufficiently clarify the scope and content of such a right. Data Protection Law and AI regulations contain more explicitly formulated rights to an explanation that also apply to certain judicial AI systems. Nevertheless, they often exclude impactful functions of judicial AI from their scope. Within these laws there is also a lack of guidance as to what explainability substantively entails. Ultimately, this patchwork of legal frameworks suggests that the protection of litigant contestation is still incomplete.

Links

Artificial intelligence, digital justice, right to an explanation

RIS

TY  - UNPD
AU  - Metikoš, L.
AU  - Iglesias Keller, C.
AU  - Qiao, C.
AU  - Helberger, N.
PY  - 2025
DA  - 2025-10-28
T1  - Comparing the Right to an Explanation of Judicial AI by Function: Studies on the EU, Brazil, and China
SP  - 31
UR  - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5674271
DO  - https://doi.org/10.2139/ssrn.5674271
AB  - Courts across the world are increasingly adopting AI to automate various tasks. But, the opacity of judicial AI systems can hinder the ability of litigants to contest vital pieces of evidence and legal observations. One proposed remedy for the inscrutability of judicial AI has been the right to an explanation. This paper provides an analysis of the scope and contents of a right to an explanation of judicial AI in the EU, Brazil, and China. We argue that such a right needs to take into account that judicial AI can perform widely different functions. We provide a classification of these functions, ranging from ancillary to impactful tasks. We subsequently compare, by function, how judicial AI would need to be explained under due process standards, Data Protection Law, and AI regulation in the EU, Brazil, and China. We find that due process standards provide a broad normative basis for a derived right to an explanation. But, these standards do not sufficiently clarify the scope and content of such a right. Data Protection Law and AI regulations contain more explicitly formulated rights to an explanation that also apply to certain judicial AI systems. Nevertheless, they often exclude impactful functions of judicial AI from their scope. Within these laws there is also a lack of guidance as to what explainability substantively entails. Ultimately, this patchwork of legal frameworks suggests that the protection of litigant contestation is still incomplete.
K1  - Artificial intelligence
K1  - digital justice
K1  - right to an explanation
ER  -

Save .RIS

Bibtex

Working paper{nokey,
	title = {Comparing the Right to an Explanation of Judicial AI by Function: Studies on the EU, Brazil, and China},
	author = {Metikoš, L. and Iglesias Keller, C. and Qiao, C. and Helberger, N.},
	doi = {https://doi.org/10.2139/ssrn.5674271},
	year = {2025},
	date = {2025-10-28},
	pages = {31},
	abstract = {Courts across the world are increasingly adopting AI to automate various tasks. But, the opacity of judicial AI systems can hinder the ability of litigants to contest vital pieces of evidence and legal observations. One proposed remedy for the inscrutability of judicial AI has been the right to an explanation. This paper provides an analysis of the scope and contents of a right to an explanation of judicial AI in the EU, Brazil, and China. We argue that such a right needs to take into account that judicial AI can perform widely different functions. We provide a classification of these functions, ranging from ancillary to impactful tasks. We subsequently compare, by function, how judicial AI would need to be explained under due process standards, Data Protection Law, and AI regulation in the EU, Brazil, and China. We find that due process standards provide a broad normative basis for a derived right to an explanation. But, these standards do not sufficiently clarify the scope and content of such a right. Data Protection Law and AI regulations contain more explicitly formulated rights to an explanation that also apply to certain judicial AI systems. Nevertheless, they often exclude impactful functions of judicial AI from their scope. Within these laws there is also a lack of guidance as to what explainability substantively entails. Ultimately, this patchwork of legal frameworks suggests that the protection of litigant contestation is still incomplete.},
	keywords = {Artificial intelligence, digital justice, right to an explanation},
}

Save .bib

Meet the Book Editor: Intellectual Property and the Human Right to a Healthy Environment external link

Izyumenko, E.

JLS Blog, 2025

Links

https://journaloflawandsociety.co.uk/blog/meet-the-book-editor-elena-izyumenko-on-intellectual-property-and-the-human-right-to-a-healthy-environment/

RIS

Save .RIS

Bibtex

Save .bib

Article 3: The Untapped Legal Basis for Europe’s Public AI Ambitions external link

Keller, P.

Links

https://legalblogs.wolterskluwer.com/copyright-blog/article-3-the-untapped-legal-basis-for-europes-public-ai-ambitions/

Artificial intelligence, CDSM Directive, Copyright, exceptions and limitations, Text and Data Mining (TDM)

RIS

Save .RIS

Bibtex

Save .bib

Top Keywords

Abstract

Links

RIS

Bibtex

Abstract

Links

RIS

Bibtex

Abstract

Links

RIS

Bibtex

Links

RIS

Bibtex

Links

RIS

Bibtex

Links

RIS

Bibtex

Links

RIS

Bibtex

Abstract

Links

RIS

Bibtex

Links

RIS

Bibtex

Links

RIS

Bibtex