three-step test

Text and Data Mining, Generative AI, and the Copyright Three-Step Test external link

IIC, vol. 57, iss. : 1, pp: 67–107, 2026

Abstract

In the debate on copyright exceptions permitting text and data mining (“TDM”) for the development of generative AI systems, the so-called “three-step test” has become a centre of gravity. The test serves as a universal yardstick for assessing the compatibility of domestic copyright exceptions with international copyright law. However, it is doubtful whether the international three-step test is applicable at all. Arguably, TDM copies fall outside the scope of the international right of reproduction and go beyond the ambit of the test’s operation. Only if national or regional copyright legislation declares the test applicable, the question arises whether copyright exceptions supporting TDM for AI training constitute certain special cases that do not conflict with a work’s normal exploitation and do not unreasonably prejudice legitimate author or rightsholder interests. As the following analysis will show, rules permitting TDM for AI training can satisfy all test criteria. An opt-out opportunity for copyright owners bans the risk of a conflict with a work’s normal exploitation and an unreasonable prejudice from the outset. A clear focus on specific policy goals, such as the objective to support scientific research, adds conceptual contours that dispel concerns about incompliance. In the case of TDM provisions covering commercial AI development, equitable remuneration regimes can be introduced as a counterbalance to avoid an unreasonable prejudice.

Links

Copyright, Generative AI, Text and Data Mining (TDM), three-step test

RIS

TY  - JOUR
AU  - Senftleben, M.
PY  - 2026
DA  - 2026-02-13
T1  - Text and Data Mining, Generative AI, and the Copyright Three-Step Test
JO  - IIC
VL  - 57
IS  - 1
SP  - 67–107
UR  - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5373903
DO  - https://doi.org/10.1007/s40319-026-01680-2
AB  - In the debate on copyright exceptions permitting text and data mining (“TDM”) for the development of generative AI systems, the so-called “three-step test” has become a centre of gravity. The test serves as a universal yardstick for assessing the compatibility of domestic copyright exceptions with international copyright law. However, it is doubtful whether the international three-step test is applicable at all. Arguably, TDM copies fall outside the scope of the international right of reproduction and go beyond the ambit of the test’s operation. Only if national or regional copyright legislation declares the test applicable, the question arises whether copyright exceptions supporting TDM for AI training constitute certain special cases that do not conflict with a work’s normal exploitation and do not unreasonably prejudice legitimate author or rightsholder interests. As the following analysis will show, rules permitting TDM for AI training can satisfy all test criteria. An opt-out opportunity for copyright owners bans the risk of a conflict with a work’s normal exploitation and an unreasonable prejudice from the outset. A clear focus on specific policy goals, such as the objective to support scientific research, adds conceptual contours that dispel concerns about incompliance. In the case of TDM provisions covering commercial AI development, equitable remuneration regimes can be introduced as a counterbalance to avoid an unreasonable prejudice.
K1  - Copyright
K1  - Generative AI
K1  - Text and Data Mining (TDM)
K1  - three-step test
ER  -

Save .RIS

Bibtex

Article{nokey,
	title = {Text and Data Mining, Generative AI, and the Copyright Three-Step Test},
	author = {Senftleben, M.},
	url = {https://link.springer.com/article/10.1007/s40319-026-01680-2},
	doi = {https://doi.org/10.1007/s40319-026-01680-2},
	year = {2026},
	date = {2026-02-13},
	journal = {IIC},
	volume = {57},
	issue = {1},
	pages = {67–107},
	abstract = {In the debate on copyright exceptions permitting text and data mining (“TDM”) for the development of generative AI systems, the so-called “three-step test” has become a centre of gravity. The test serves as a universal yardstick for assessing the compatibility of domestic copyright exceptions with international copyright law. However, it is doubtful whether the international three-step test is applicable at all. Arguably, TDM copies fall outside the scope of the international right of reproduction and go beyond the ambit of the test’s operation. Only if national or regional copyright legislation declares the test applicable, the question arises whether copyright exceptions supporting TDM for AI training constitute certain special cases that do not conflict with a work’s normal exploitation and do not unreasonably prejudice legitimate author or rightsholder interests. As the following analysis will show, rules permitting TDM for AI training can satisfy all test criteria. An opt-out opportunity for copyright owners bans the risk of a conflict with a work’s normal exploitation and an unreasonable prejudice from the outset. A clear focus on specific policy goals, such as the objective to support scientific research, adds conceptual contours that dispel concerns about incompliance. In the case of TDM provisions covering commercial AI development, equitable remuneration regimes can be introduced as a counterbalance to avoid an unreasonable prejudice.},
	keywords = {Copyright, Generative AI, Text and Data Mining (TDM), three-step test},
}

Save .bib

Are the European TDM Exceptions Applicable to GenAI Training? Despite the Three-Step Test? external link

Senftleben, M.

Kluwer Copyright Blog, 2025

Copyright, GenAI, Text and Data Mining (TDM), three-step test

RIS

Save .RIS

Bibtex

Save .bib

GenAI and the Copyright Three-Step Test – Do TDM Exceptions for AI Training Conflict With a Work’s Normal Exploitation? external link

Senftleben, M.

GRUR International, vol. 75, iss. : 1, pp: 1-2, 2025

Abstract

Text and data mining (TDM) for AI training can be regarded as the starting point of a complex process that impacts the market for human literary and artistic creations in different ways. The machine is only capable of mimicking human content after it had the opportunity to derive patterns for its own productions from myriad human creations that served as training resources. Once AI training has been completed and a generative AI (GenAI) system is brought to the market, AI output may support fruitful human/machine collaboration. However, it may also kill demand for the same human creativity that empowered the AI system to become a competitor in the first place. In the terminology of the ubiquitous three-step test in international and European copyright law, this latter challenge raises the question whether copyright exceptions permitting TDM for AI training cause a conflict with a work’s normal exploitation. A closer inspection of the normal exploitation test shows that the chances of demonstrating a relevant conflict are slim in the case of AI training. Rightsholders seeking compensation for displacement effects caused by GenAI systems must resort to the final criterion of the three-step test and argue that the use for AI development unreasonably prejudices their legitimate interests. In practice, this means that copyright holders can hardly employ the three-step test as a tool to erode TDM exemptions altogether. They can only insist on the introduction of appropriate remuneration schemes to avoid unreasonable prejudice in cases of commercial AI training.

Links

DOI: https://doi.org/10.1093/grurint/ikaf113

Copyright, exploitation, GenAI, Text and Data Mining (TDM), three-step test

RIS

TY  - JOUR
AU  - Senftleben, M.
PY  - 2025
DA  - 2025-08-29
T1  - GenAI and the Copyright Three-Step Test – Do TDM Exceptions for AI Training Conflict With a Work’s Normal Exploitation?
JO  - GRUR International
VL  - 75
IS  - 1
SP  - 1-2
UR  - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5356851
DO  - https://doi.org/10.1093/grurint/ikaf113
AB  - Text and data mining (TDM) for AI training can be regarded as the starting point of a complex process that impacts the market for human literary and artistic creations in different ways. The machine is only capable of mimicking human content after it had the opportunity to derive patterns for its own productions from myriad human creations that served as training resources. Once AI training has been completed and a generative AI (GenAI) system is brought to the market, AI output may support fruitful human/machine collaboration. However, it may also kill demand for the same human creativity that empowered the AI system to become a competitor in the first place. In the terminology of the ubiquitous three-step test in international and European copyright law, this latter challenge raises the question whether copyright exceptions permitting TDM for AI training cause a conflict with a work’s normal exploitation.

A closer inspection of the normal exploitation test shows that the chances of demonstrating a relevant conflict are slim in the case of AI training. Rightsholders seeking compensation for displacement effects caused by GenAI systems must resort to the final criterion of the three-step test and argue that the use for AI development unreasonably prejudices their legitimate interests. In practice, this means that copyright holders can hardly employ the three-step test as a tool to erode TDM exemptions altogether. They can only insist on the introduction of appropriate remuneration schemes to avoid unreasonable prejudice in cases of commercial AI training.
K1  - Copyright
K1  - exploitation
K1  - GenAI
K1  - Text and Data Mining (TDM)
K1  - three-step test
ER  -

Save .RIS

Bibtex

Article{nokey,
	title = {GenAI and the Copyright Three-Step Test – Do TDM Exceptions for AI Training Conflict With a Work’s Normal Exploitation?},
	author = {Senftleben, M.},
	doi = {https://doi.org/10.1093/grurint/ikaf113},
	year = {2025},
	date = {2025-08-29},
	journal = {GRUR International},
	volume = {75},
	issue = {1},
	pages = {1-2},
	abstract = {Text and data mining (TDM) for AI training can be regarded as the starting point of a complex process that impacts the market for human literary and artistic creations in different ways. The machine is only capable of mimicking human content after it had the opportunity to derive patterns for its own productions from myriad human creations that served as training resources. Once AI training has been completed and a generative AI (GenAI) system is brought to the market, AI output may support fruitful human/machine collaboration. However, it may also kill demand for the same human creativity that empowered the AI system to become a competitor in the first place. In the terminology of the ubiquitous three-step test in international and European copyright law, this latter challenge raises the question whether copyright exceptions permitting TDM for AI training cause a conflict with a work’s normal exploitation.

A closer inspection of the normal exploitation test shows that the chances of demonstrating a relevant conflict are slim in the case of AI training. Rightsholders seeking compensation for displacement effects caused by GenAI systems must resort to the final criterion of the three-step test and argue that the use for AI development unreasonably prejudices their legitimate interests. In practice, this means that copyright holders can hardly employ the three-step test as a tool to erode TDM exemptions altogether. They can only insist on the introduction of appropriate remuneration schemes to avoid unreasonable prejudice in cases of commercial AI training.},
	keywords = {Copyright, exploitation, GenAI, Text and Data Mining (TDM), three-step test},
}

Save .bib

Towards a European Research Freedom Act: A Reform Agenda for Research Exceptions in the EU Copyright Acquis external link

Senftleben, M., Szkalej, K., Sganga, C. & Margoni, T.

IIC, vol. 56, iss. : 7, pp: 1329–1358, 2025

Abstract

This article explores the impact of EU copyright law on the use of protected knowledge resources in scientific research contexts. Surveying the current copyright/research interface, it becomes apparent that the existing legal framework fails to offer adequate balancing tools for the reconciliation of divergent interests of copyright holders and researchers. The analysis identifies structural deficiencies, such as fragmented and overly restrictive research exceptions, opaque lawful access provisions, outdated non-commercial use requirements, legal uncertainty arising from the three-step test in the EU copyright acquis, obstacles posed by the protection of paywalls and other technological measures, and exposure to contracts that override statutory research freedoms. Empirical data confirm that access barriers, use restrictions and the absence of harmonised rules for transnational research collaborations impede the work of researchers. Against this background, we advance proposals for legislative reform, in particular the introduction of a mandatory, open-ended research exemption that offers reliable breathing space for scientific research across EU Member States, the clarification of lawful access criteria, a more flexible approach to public-private partnerships, and additional rules that support modern research methods, such as text and data mining.

Links

Copyright, open science, research exceptions, right to research, technological protection measures, Text and Data Mining (TDM), three-step test

RIS

TY  - JOUR
AU  - Senftleben, M.
AU  - Szkalej, K.
AU  - Sganga, C.
AU  - Margoni, T.
PY  - 2025
DA  - 2025-07-22
T1  - Towards a European Research Freedom Act: A Reform Agenda for Research Exceptions in the EU Copyright Acquis
JO  - IIC
VL  - 56
IS  - 7
SP  - 1329–1358
UR  - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5130069
DO  - https://doi.org/10.1007/s40319-025-01604-6
AB  - This article explores the impact of EU copyright law on the use of protected knowledge resources in scientific research contexts. Surveying the current copyright/research interface, it becomes apparent that the existing legal framework fails to offer adequate balancing tools for the reconciliation of divergent interests of copyright holders and researchers. The analysis identifies structural deficiencies, such as fragmented and overly restrictive research exceptions, opaque lawful access provisions, outdated non-commercial use requirements, legal uncertainty arising from the three-step test in the EU copyright acquis, obstacles posed by the protection of paywalls and other technological measures, and exposure to contracts that override statutory research freedoms. Empirical data confirm that access barriers, use restrictions and the absence of harmonised rules for transnational research collaborations impede the work of researchers. Against this background, we advance proposals for legislative reform, in particular the introduction of a mandatory, open-ended research exemption that offers reliable breathing space for scientific research across EU Member States, the clarification of lawful access criteria, a more flexible approach to public-private partnerships, and additional rules that support modern research methods, such as text and data mining.
K1  - Copyright
K1  - open science
K1  - research exceptions
K1  - right to research
K1  - technological protection measures
K1  - Text and Data Mining (TDM)
K1  - three-step test
ER  -

Save .RIS

Bibtex

Article{nokey,
	title = {Towards a European Research Freedom Act: A Reform Agenda for Research Exceptions in the EU Copyright Acquis},
	author = {Senftleben, M. and Szkalej, K. and Sganga, C. and Margoni, T.},
	url = {https://link.springer.com/article/10.1007/s40319-025-01604-6},
	doi = {https://doi.org/10.1007/s40319-025-01604-6},
	year = {2025},
	date = {2025-07-22},
	journal = {IIC},
	volume = {56},
	issue = {7},
	pages = {1329–1358},
	abstract = {This article explores the impact of EU copyright law on the use of protected knowledge resources in scientific research contexts. Surveying the current copyright/research interface, it becomes apparent that the existing legal framework fails to offer adequate balancing tools for the reconciliation of divergent interests of copyright holders and researchers. The analysis identifies structural deficiencies, such as fragmented and overly restrictive research exceptions, opaque lawful access provisions, outdated non-commercial use requirements, legal uncertainty arising from the three-step test in the EU copyright acquis, obstacles posed by the protection of paywalls and other technological measures, and exposure to contracts that override statutory research freedoms. Empirical data confirm that access barriers, use restrictions and the absence of harmonised rules for transnational research collaborations impede the work of researchers. Against this background, we advance proposals for legislative reform, in particular the introduction of a mandatory, open-ended research exemption that offers reliable breathing space for scientific research across EU Member States, the clarification of lawful access criteria, a more flexible approach to public-private partnerships, and additional rules that support modern research methods, such as text and data mining.},
	keywords = {Copyright, open science, research exceptions, right to research, technological protection measures, Text and Data Mining (TDM), three-step test},
}

Save .bib

Generative AI and Author Remuneration

Senftleben, M.

IIC, vol. 54, pp: 1535-1560, 2023

Abstract

With the evolution of generative AI systems, machine-made productions in the literary and artistic field have reached a level of refinement that allows them to replace human creations. The increasing sophistication of AI systems will inevitably disrupt the market for human literary and artistic works. Generative AI systems provide literary and artistic output much faster and cheaper. It is therefore foreseeable that human authors will be exposed to substitution effects. They may lose income as they are replaced by machines in sectors ranging from journalism and writing to music and visual arts. Considering this trend, the question arises whether it is advisable to take measures to compensate human authors for the reduction in their market share and income. Copyright law could serve as a tool to introduce an AI levy system and ensure the payment of equitable remuneration. In combination with mandatory collective rights management, the new revenue stream could be used to finance social and cultural funds that improve the working and living conditions of flesh-and-blood authors.

Links

DOI: https://doi.org/10.1007/s40319-023-01399-4

collective rights management, Copyright, Freedom of expression, Text and Data Mining (TDM), three-step test

RIS

TY  - JOUR
AU  - Senftleben, M.
PY  - 2023
DA  - 2023-11-07
T1  - Generative AI and Author Remuneration
JO  - IIC
VL  - 54
SP  - 1535-1560
DO  - https://doi.org/10.1007/s40319-023-01399-4
AB  - With the evolution of generative AI systems, machine-made productions in the literary and artistic field have reached a level of refinement that allows them to replace human creations. The increasing sophistication of AI systems will inevitably disrupt the market for human literary and artistic works. Generative AI systems provide literary and artistic output much faster and cheaper. It is therefore foreseeable that human authors will be exposed to substitution effects. They may lose income as they are replaced by machines in sectors ranging from journalism and writing to music and visual arts. Considering this trend, the question arises whether it is advisable to take measures to compensate human authors for the reduction in their market share and income. Copyright law could serve as a tool to introduce an AI levy system and ensure the payment of equitable remuneration. In combination with mandatory collective rights management, the new revenue stream could be used to finance social and cultural funds that improve the working and living conditions of flesh-and-blood authors.
K1  - collective rights management
K1  - Copyright
K1  - Freedom of expression
K1  - Text and Data Mining (TDM)
K1  - three-step test
ER  -

Save .RIS

Bibtex

Article{nokey,
	title = {Generative AI and Author Remuneration},
	author = {Senftleben, M.},
	doi = {https://doi.org/10.1007/s40319-023-01399-4},
	year = {2023},
	date = {2023-11-07},
	journal = {IIC},
	volume = {54},
	pages = {1535-1560},
	abstract = {With the evolution of generative AI systems, machine-made productions in the literary and artistic field have reached a level of refinement that allows them to replace human creations. The increasing sophistication of AI systems will inevitably disrupt the market for human literary and artistic works. Generative AI systems provide literary and artistic output much faster and cheaper. It is therefore foreseeable that human authors will be exposed to substitution effects. They may lose income as they are replaced by machines in sectors ranging from journalism and writing to music and visual arts. Considering this trend, the question arises whether it is advisable to take measures to compensate human authors for the reduction in their market share and income. Copyright law could serve as a tool to introduce an AI levy system and ensure the payment of equitable remuneration. In combination with mandatory collective rights management, the new revenue stream could be used to finance social and cultural funds that improve the working and living conditions of flesh-and-blood authors.},
	keywords = {collective rights management, Copyright, Freedom of expression, Text and Data Mining (TDM), three-step test},
}

Save .bib