Beteiligung zum Fragebogen der EU-Kommission: Zur Wirksamkeit und zu den Auswirkungen der DSM-Richtlinie in den Mitgliedstaaten

-Konsultationsantwort des Netzwerk Autorenrechte (NAR)-

-Präambel-

Wir bedanken uns für die Einladung des BMJV Ref III B3, überaus kurzfristig zur Evaluation der Europäischen Kommission der Richtlinie 2019/790 (EU), „CDSM Directive“, Stellung zu nehmen. Unsere Antworten werden auf Englisch erfolgen, die Einleitung auf Deutsch.

ZUM DOWNLOAD

Das Netzwerk Autorenrechte (NAR; www.netzwerk-autorenrechte.de) vereint 15 Verbände aus Deutschland, Österreich und der Schweiz und repräsentiert 16.000 Autor:innen und Übersetzer:innen des deutschsprachigen Buchsektors. Die Mitglieder der Verbände schreiben, übersetzen und veröffentlichen in allen Gattungen und Genres.

Das Netzwerk Autorenrechte ist darüber hinaus vertreten im European Writers‘ Council sowie PLR  International und repräsentiert die Interessen der Urheber:innen des Buchsektors gegenüber dem Europaparlament und seinen Ausschüssen, der Europäischen Kommission, der WIPO, der UNESCO, war u.a. über seine individuellen politischen Beauftragten seit 2014 an der Entwicklung der UR-RL beteiligt und begleitete später deren Implementation in Deutschland sowie den Aufbau des OOCW-Portals bei der EUIPO. Mitglieder des NAR sind Verwaltungsräte der VG Wort und begleiteten u.a. die Aushandlung der kollektiven Lizenzen unter UR-RL Art 8.

Besonderes Augenmerk legten wir schon in der Referententwurfphase auf zwei Sollbruchstellen:

• Die dysfunktionale Implementierung der urhebervertragsrechtlichen Artikel (18-23) durch die Unterschlagung einer „verhältnismäßigen“ gesetzlichen Vergütung sowie der Abschwächung der Auskunftspflicht dritter Vertragsparteien in ein – bis heute nichtdurchsetzbares – Auskunftsrecht;
• Die Gefahr der technischen und damit auch rechtlichen Überinterpretation der TDM-Ausnahmen (3,4), die eben nicht für die Entwicklung generativer Systeme und Modelle sogenannter „KI“ gedacht waren.

Wir weisen zudem auf folgende europaweite Erhebungen hin, die wir das BMJV bitten, dem Konsortium, das im Auftrag der KOM die Evaluation begleitet, als Quellen zukommen zu lassen:

• Contracts, Remuneration and Transparency: A comprehensive Report on Contractual Clauses in Publishing Agreements in the European Book Sector under the Directive 2019/790 (EU) (European Writers’ Council, 2024), https://europeanwriterscouncil.eu/writers-contracts-ewcsurvey-2024/
• Well intentioned – but well implemented? (EWC Survey, 2022) https://europeanwriterscouncil.eu/cdsm_surveyresults2022/

Berlin, 20. März 2026. Für das Netzwerk Autorenrechte:

Nina George, politische Beauftragte NAR; EWC, nina.george(at)europeanwriterscouncil.eu

 

I.          Impact of the DSM Directive on the digital use of protected content for education, research and preservation (Articles 3-7)  

I.1. Article 3 – Text and data mining for the purposes of scientific research […]

4. Does the national implementation of the TDM exception introduced in Article 3 facilitate cross-border research collaboration, and partnerships between research organisations and industry? If any, please indicate any legal or practical barriers encountered in this context.
5. In your view, to what extent has the TDM exception introduced in Article 3 been effective in your Member State in improving the ability of research organisations and cultural heritage institutions to carry out text and data mining for scientific research, including in the context of AI development? In your response, please consider whether the implementation has increased legal certainty for eligible institutions, improved practical access to content for TDM, reduced technical or contractual obstacles to conducting TDM, and led to any increase or change in TDM activities.

RESPONSE to Q4 and Q5

a) The Netzwerk Autorenrechte (NAR; Network Authors’ Rights) responds with the proviso that the premise of this consultation needs formal clarification. The exception(s) for text and data mining (Art. 3, Art. 4, 2019/790 (EU)) are inapplicable with respect to AI and GenAI development: The statutory language and text of the provision(s), its conception, and the ratio of the exception esp. Art 4. indicate that it must not be applied to the training of any (generative) AI models. Hence, without the authors’ authorisation any use by commercial developers is to be classified as copyright infringement.

This also applies to authors whose works are used in research and academia beyond text and data mining (TDM) under the non-commercial exception – and to any private partnership between an academic institution and a private-sector company that unlawfully exploits collections of works for commercial AI development.

• TDM is not equal to AI: this has been examined and confirmed on several occasions since 2021 by various international experts, most recently by Dr Nicola Lucchi, among others, on behalf of the JURI Committee[1],[2].
• The resolution[3] adopted by the European Parliament on 10 March by a large majority highlights on five separate occasions the flaw in the misinterpretation of the TDM exception and the fact that TDM is not the same as AI development.
• Here, the Commission should follow the elected representatives of the EP, who designed the 2019/790, have also repeatedly stated publicly[4] and with political impact over the past few years that the exceptions were not intended to provide AI developers with a freewheeling buffet in Europe.

b) Germany already had a TDM exemption that was subject to remuneration (organised collectively through CMOs), and the framework for which clearly ruled out any use for the development of advanced technologies. Background: On 1 March 2018, the revised copyright law for the academic sector (the Copyright and Knowledge Society Act; UrhWissG) came into force in Germany, amending the previous Copyright Act and regulating the use of copyright-protected works in the digital age; this included text and data mining for research and teaching (UrhG 60d). Section 60d of the German Copyright Act (UrhG) first regulated so-called text and data mining, and made clear, that TDM is “the automated analysis of large volumes of copyright-protected content” for non-commercial purposes. Such use was subject to a compensation fee and the copies of the works had to be deleted after use but could be transferred to a library or archive for preservation purposes. 

Since the introduction of Articles 2 and 3 of Directive 2019/790, numerous universities and research institutions have passed on the text corpora they use as datasets to commercial companies that provide financial support to these institutions.

This results in the following serious consequences for the affected owners of the authors rights and the copyright:

• whereas rights holders can not opt out of the text and data mining carried out in their catalogues under this Article 3, as they are most often not made aware that their works are included in corpora shared with a commercial entity;
• This constitutes a breach of the Directive, as the use of works may initially be regarded as falling under Article 3, but subsequently under Article 4, which should allow for the possibility of a reservation of rights. Accordingly, it is our view that the thousands of corpora of works passed on by research organisations to commercial entities cannot under any circumstances be considered ‘legally’ or ‘publicly’ available and accessible.

As the more than 100 lawsuits worldwide show, authors and AI developers do not have in general the same understanding of what exactly ‘legally accessible’ and ‘publicly available’ mean.

Often, AI developers work with data suppliers, who in turn make use of works beyond paywalls. It is unclear here whether opt-outs are respected, whether works were purchased, when they were purchased and reproduced, whether a reservation of rights was recognised and accepted in, for example, the legal notice, etc.

Accordingly, the transparency chain must start from the moment Zero of scrawling, and also consider cases in which works have been collected under Art 3 TDM exception at a research institute, but then passed on in private partnership with commercial AI providers.

This is neither legal nor public, but it leaves academic authors, among others, severely damaged, who would also have to be allowed to opt out when switching from one exception to the other.

We also point to cases where, for example, videos have been transcribed into text and sold or transferred as a ‘dataset’; these cases also make a reservation of rights impossible, as well as tracking in order to substantiate one’s rights in court proceedings.

In addition, reliable and fine-grained definitions need to take place, also to give authors and rights holders the securitised option to enforce their rights, and to start the transparency chain already with the collectors and curators, confirming legal access. 

I.2. Article 4 – Exception or limitation for text and data mining

7. To what extent has your national implementation of the Article 4 TDM exception been effective in improving the ability to carry out text and data mining, including in the context of AI? Are you aware of specific situations / use cases in which the exception has been used?  Please specify.
8. Has the national implementation of Article 4 enabled text and data mining in areas beyond large language model development? Please indicate concrete cases or use cases you are aware of, and any obstacles encountered.

RESPONSE to Q6 and Q7

The Netzwerk Autorenrechte (NAR; Network Authors’ Rights) responds with the proviso that the premise of this consultation needs formal clarification. The exception(s) for text and data mining (Art. 3, Art. 4, 2019/790 (EU)) are inapplicable with respect to AI and GenAI development: The statutory language and text of the provision(s), its conception, and the ratio of the exception esp. Art 4. indicate that it must not be applied to the training of any generative AI models. Hence, without the authors’ authorisation any use by commercial developers is to be classified as copyright infringement.  

• This has been examined and confirmed on several occasions since 2021 by various international experts, most recently by Dr Nicola Lucchi, among others, on behalf of the JURI Committee[5],[6].
• The resolution[7] adopted by the European Parliament on 10 March by a large majority highlights on five separate occasions the flaw in the misinterpretation of the TDM exception and the fact that TDM is not the same as AI development.
• Here, the Commission should follow the elected representatives of the EP, who designed the 2019/790, have also repeatedly stated publicly[8] and with political impact over the past few years that the exceptions were not intended to provide commercial LLMs and other AI developers with a freewheeling buffet in Europe.

Technical steps of TDM in comparison to (gen)AI and their legal relations:

• During the training of a generative AI model, several acts of copying and reproduction of copyright-protected works within the meaning of the InfoSoc Directive (EU) occur:
• (1) It starts with the collection, preparation, and storage of copyrighted works used for the (Gen)AI training process. Each new cycle of an AI system and AI model is relying on stored works to use again. The TDM exception instead mandates to a deleting of copies.
• (2) In addition, during pre-development and fine-tuning, copyright-relevant reproductions of copyrighted works materialise “inside” the AI model. This also constitutes a copy and replication in the legal sense.
• (3) Furthermore, during the application of generative AI models, particularly by the end users of the fully trained AI systems (e.g., ChatGPT via the OpenAI website), works that have been used for training the AI model may be copied and replicated as part of the systems’ output. This had been proven by several lawsuits in Germany (GEMA) but also around the world.
• (4) Finally, what has been overlooked so far, the making available of generative AI models that have been implemented in AI systems for the users of these systems (again, ChatGPT via the OpenAI website) or for downloading as a whole constitutes a making available to the public of the works replicated “inside” the generative AI models according to Article 3 of the InfoSoc Directive.

 

NAR’s position is currently to recommend that the TDM reservation of rights (‘opt-out’) be declared a precautionary measure to avoid the misuse of any works for generative AI (GenAI), but, at the same time, we support the correct legal perspective thatexploitation for the development of generative AI is a new exclusive right not covered by both the TDM exemptions. We are employing the opt-out as a protective shield, not as acceptance of the overinterpretation of the TDM exceptions.

TDMRep has a dramatic disadvantage for authors: it is inaccessible to them as they are the owners of the rights, i.e. TDM, AI, generative AI. This means that those who are uniquely entitled to opt out are unable to technically reserve their rights. As a result, authors are dependent on publishers, intermediaries or distributors, and increasingly being forced to transfer exclusive rights in order to opt-out – and stay helpless. The TDMRep is not recognised as an internationally standard.

Contractual overrun of legal provisions under CC-Licences

Authors as originators of all works are the primary owners of moral rights and of any rights related to advanced informatics, incl. for TDM, for AI, and for GenAI development. None of these rights had been contractually transferred in the consumer book market but started to be forced to be transferred within expanded CC licences in the academic book sphere, which makes the individual opt-out under TDM Art. 4 (2019/790) impossible.

Specific use cases:

German writers and translators of the book sector are among the most ripped-off independent cultural workers by mostly non-European technology companies, which have developed highly profitable programs based on our works and books, created at our own financial risk. This does not only relate to LLM, but also “AI features”, which normally would fall under licensing obligations; e.g. producing summaries, recaps, interactive enhanced e-books with built-in chatbots.

Further, the works and data exploiting does not stop with LLM; images, including children’s and minor’s sexualizing images, in addition voices, appearances, faces, are “mined” to be reproduced in diffusion models or systems reproducing works in composition and music.

The several breaches under the overinterpretation of the TDM exceptions also touch upon the violation of GDPR and the right to personal information.

We focus in our further response on specific use cases around text robotics.

There are currently over 1,000 large language models in use worldwide, dozens of which have a computing capacity of 23^10 FLOP and a database of several million protected text works, often books published well before 2021 and the entry into force of the CDSM Directive 2019/790 (EU). These generative models (re)produce text and translative transfers, the “machine translation”.

The analysis the Network Authors Rights (NAR) in 2023[9] (translation into English and re-published at the EWC) already showed, how the promised success of “AI” is built on a widely over-interpretation of the non-remunerated TDM exception of Art. 4, and led to a massive exploitation of labour hours, private investment, context, and works, and how the output and books-imitating products harm remuneration, reputation, and reinforces bias, language discrimination, and leads to heavy de-skilling.

Over the past three years, translators in the five larger markets including Germany saw a decline of commissions by a third, illustrators by a fourth. Authors are competing with AI-outputs on their market, and are hindered to exploit their works normally, have a loss of licensing revenues in the billions, while AI-products simulating books enjoy privileges meant for human creation–for example CMO revenues, prizes, scholarships, and other benefits meant for cultural assets by human originality.

Selected examples / use cases:

(a) Uncontrolled AI output is being pushed into the bestseller lists with click farms: For years, the global provider Amazon has been flooded with thousands of thousands bogus books by fake authors whose text and visual products have been produced by generative text and image output software. AI bots from click farms “read” these book-imitating works, articulate random machine-made reviews, and push them into the bestseller lists[10]. This led to a rapid decline in revenue for human authors by shared-revenue models, such as Kindle KDP (A pool of revenue divided by pages read and number of authors, similar to Spotify). At peak times, 80 out of 100 Kindle KDP bestsellers are AI editions. Amazon rejects to label AI output, although

(b) Identity theft and name deception: The world’s most important review platform Goodreads, like Amazon, is flooded with AI books published under the illegitimately used names of real human authors (or slightly altered spellings of known writers). These books are listed as new releases in the authors’ profiles and entice readers to buy them. However, the income from these AI books flows to
unknown sources. Human authors who are cheated out of their earnings must spend money to defend themselves with lawyers. So far, neither Goodreads nor Amazon have stopped this identity theft, which damages the reputation of human authors. 

(c) Unauthorised machine translations open up foreign-language markets and channel sales to unknown sources: We observed cases of books being illegally translated from, for example, the English language into Spanish and Portuguese by means of robot translation without a license, and published under a different name, usually in Amazon Selfpublishing. The revenues flow to unknown sources.

(d) Illegal remuneration claims to collective management organisations and media clients: It cannot be ruled out that automatically generated and machine translated press articles and machine translated books, or even regenerative produced AI images, already “enjoy” private copying remuneration from collective managements organisations (CMOs), as there is no legal labelling obligation yet; or generated texts, machine translations and generated images flow into the media on a royalty basis.

Likewise, scholarships, prizes or State’s grants are–Unintentionally–provided towards AI products instead incentivising human works.

(f) Amazon is using AI for non-licensed derivates of books or enhanced AI-features

The “Recaps” (“Story So Far”) feature summarising prior books. How Amazon approached acquiring rights to copy and ingest full books in order to create AI-generated summaries, is until to date not answered by this non-European monopoly.

Kindle Translate is an AI-based translation, launched 6/11/25, offering machine translation from English and Spanish, and from German to English. This will impact translators’ job massively.

The feature Ask this Book is since 11/12/25 available for “thousands of English-language books” on the Kindle iOS app in the U.S. and let readers ask questions about the e-books they are reading.  Requests on licensing for this feature or elaborate on the technical details of the service–e-.g. if the book was fed into the in-house AI-LLM– and any protections involved (whether to prevent against hallucinations, or to protect the text from further AI training), are still not answered by the Non-European market giant; who seemed also to ignore a proper TDM opt-out.

To use chat-bots and summaries on a work and book, a derivative work is created and turns books into searchable, interactive products akin to enhanced e-books or annotated editions—a new format for which rights should be negotiated. Likewise, in creating a chat feature that allows readers to ask questions about a book—including analysis and summaries—Amazon is creating a derivate use, which also harms remuneration and reputation of the authors. Amazon’s unilateral, unlicensed, no-opt-out incorporation these AI-driven features sets a dangerous precedent.

• Overall: the impact on collateral consequences, including the States and European societies, e.g. fewer payments of taxes, into pension schemes or social insurances systems and further societal and culture-economic aspects, are side effects of the misinterpretation of the exception.
• We call onto the Commission to cancel both the exceptions.

I.3. Article 5 – Use of works and other subject matters in digital and cross-border teaching activities

[…]

10. To what extent has your national implementation of Article 5 been effective in improving the use of works and other subject matter in digital and cross-border online teaching activities?

– no response available- 

I.4. Article 6 – Preservation of cultural heritage

11. To what extent has your national implementation of Article 6 improved the ability of cultural heritage institutions (CHIs) to preserve works or other subject matter? In your response, please consider the following: practical preservation activities enabled by the Article 6 exception (e.g., digitisation, archiving, format-shifting, overcoming technological protection measures); any observable increase in preserved content; examples of successful preservation projects undertaken by CHIs since implementation; whether the implementation reduced legal, technical, or organisational barriers to preservation.

– no response available-

I.5. Cross-cutting implementation observations for articles 3 – 7

[…]

– no response available-

II. Impact of the DSM Directive on licensing practices and the online access to protected content across the EU (Articles 8-14)

II.1. Articles 8 and 9 – Out-of-commerce works and cross-border uses

15. To what extent has your national implementation of Article 8 been effective in facilitating licensing practices for out-of-commerce works (OOCWs) and improving access to such works by cultural heritage institutions? Please consider whether the implementation has enabled the conclusion of licences in practice and led to wider availability or use of out-of-commerce works.

RESPONSE to Q15

To address the issue of out-of-print / out of commerce works in published publications, the German collective rights management organisations VG WORT and VG Bild-Kunst have concluded a framework agreement with the federal and state governments regarding the use of such works. The granting of rights is conditional upon accession to the framework agreement.

A work is considered unavailable if it is not offered to the general public in its complete form through any standard distribution channel. The date of the last publication must be at least 30 years ago (moving time wall).

• The licensing offer is provided in cooperation with the German National Library (DNB). Works are registered via the DNB; the framework agreement and the declaration of accession are sent upon submission of the first registration.
• VG WORT handles the accounting and licensing in accordance with the tariffs specified in the framework agreement, as well as the distribution of the amounts to the rights holders (authors and publishers).
• Certain information about the work must be published by VG WORT six months prior to the granting of rights on the Out-of-commerce works Portal (OOCW) of the European Union Intellectual Property Office (EUIPO ). Rights holders may object to the use of the work via the portal at any time, including after the six-month period has expired. In the event of a valid objection, no further use or distribution may take place.

We note that the EUIPO OOCW portal remains non-functional and is neither adequately prepared for the transfer of catalogues nor for the exercise of reservations of rights by authors. Authors are the original rights holders, as when a work becomes OOCW, the rights revert to the author. Accordingly, they must be able to express their rights preferences effectively, easily and without technical difficulty.

Already in 2019, our umbrella organisation, the EWC, laid down the necessary steps to a functional application of Art. 8 / 2019/790 (EU) from the perspective of the affected rightsholder, the authors:

Authors and out-of-commerce-works: conditions of exploitation under Art 8-11

• All contracts (in the EU) or egal provisions needs to have a clear revocation rights clause in the contract (see DSM, Article 22: Right of revocation; in Germany with a diffuse implementation);
• must have: a clear definition of OOCW; this might be a moving time wall (Germany has set this up: 30 years) AND a clarification, that OOCW always is related to print editions, not to e-books.
• must have: clear proceedings of reversion of rights and/or clearing methods, if a publisher also claims to still have transferred primary rights;
• must have: the author should be informed before the work is digitised and distributed, as it should not be on the burden of the author to frequently research for their book;
• An author needs to be informed about the registry and is allowed opt-out at any time with no time-window
• In addition: transparency and frequent overviews of usage-data leads to the chance to pay appropriate and proportionate remuneration, not only compensation;
• Any usage of digitized OOCW shall take place in a closed garden / intranet; no download, no print-out
• the work should not be modified, altered, adopted or abridged in any way, to not breach the right to integrity of the author onto their works;
• And most important: CHI are not allowed to grant any AI developer or data collector access to these OOCW, to avoid misuse again. If they do so, they shall inform the author and manage the opt out declarations.

It is precisely on this last point that new misunderstandings have arisen in Germany, and indeed across Europe: for instance, cultural heritage institutions (CHI), including those operating under the Europeana umbrella, assume that works digitised and made accessible under Article 8 (OOCW) may be transferred to commercial AI developers as collections of works and book corpora.

This is, of course, illegal and in no way permitted by the law or the correct interpretation of the Directive – yet, in purely practical terms, it has already led on a massive scale to CHIs making all OOCW publicly accessible via websites, copyable and without any indication that this practice is illegal – and AI developers are also making diligent use of these OOCW works for the commercial production of generative and other so-called ‘AI’. Historical images, gallery and museum catalogues, books, musical works, song lyrics, sheet music – a gigantic quarry, and not a single opportunity for authors to reserve their rights regarding TDM use, or to say no outright to AI use, or to demand licensing.

A massive loophole has been created here, which makes the greatest theft in human history – of creative works and investments in art and knowledge – all the more outrageous.

In Germany, an agreement was reached within the framework agreement (see the introductory comment on the question) which excludes TDM under Article 4 and provides for OOCW works to be entered into a register with an opt-out. However, this does not apply to works that could fall under Article 3; this is all the more grave given that academic authors whose works are OOCW are deprived twice over of their legitimate rights to object when their subsequently digitised OOCW are handed over to commercial AI developers.

17. Do licences concluded under your national implementation of Article 8 in practice include cross-border use to make out-of-commerce works available for cross-border access in accordance with Article 9?

-please consult the response by VG Wort-

18. Are you aware if in practice cultural heritage institutions rely on the exception under Article 8(2) for making available OOCWs? If so, in which situations?

-please consult the response by VG Wort- 

II.3. Article 11 – Stakeholder dialogue

II. 4. Article 12 – Collective licensing with an extended effect

21. […] If applicable, to what extent have the rules on extended collective licensing been effective in facilitating access or use of copyright protected content? Are you aware of new licensing agreements concluded under these rules since the transposition of the DSM Directive?

 -please consult the response by VG Wort-

II.5. Article 13 – Video-on-demand negotiation mechanism

[…]

23. Is any information available on the extent to which this mechanism has been relied upon or how its use is encouraged? To the best of your knowledge, how has the mechanism addressed difficulties in the acquisition and clearance of rights?

– no response available-

II.6. Article 14 – Works of visual art in the public domain 

25. To what extent has your national implementation of Article 14 facilitated the use of nonoriginal reproductions of works of visual art which are in the public domain?

– no response available- 

II.7. Cross-cutting implementation observations for articles 8 – 14

If a CHI gives access to a dataset of digitised OoC books to train TDM/(G)AI, it will deny authors from their right to opt out of such TDM or further usage for (G)AI. The effectiveness  of the opt out would no longer be possible. But more and more CHI misuse the Art 8. To act under Art 4.

Legal considerations:

1. The license can only be granted for non-commercial purposes (Article 8). TDM under Article 4 would apply to companies that are not defined as research organisations (which are covered by Article 3), thus breaking the conditions of the license. Recital 13: Conversely, organisations upon which commercial undertakings have a decisive influence allowing such undertakings to exercise control because of structural situations, such as through their quality of shareholder or member, which could result in preferential access to the results of the research, should not be considered research organisations for the purposes of this Directive.

2.Books that have been digitised and made available under the OoCW license cannot be considered as lawfully accessible to be used for TDM/(G)AI, unless the rights holders have explicit permitted these uses.

3. Additionally, even if such commercial uses were permitted by the author, it could not claim the benefit of an ECL under Article 12, as those under Article 8 can only cover non-commercial uses and nothing else. Therefore, in such a case, only the authors who explicitly agreed to have their content mined for commercial purposes could be covered by this license.

4. The exception under Article 4 as defined in Recital 18 should only apply where the work or other subject matter is accessed lawfully by the beneficiary, including when it has been made available to the public online, and insofar as the authors have not reserved in an appropriate manner the rights to make reproductions and extractions for text and data mining. In the case of works / books that has been made publicly available online, it should only be considered appropriate to reserve those rights by the use of machine-readable means, including metadata and terms and conditions of a website or a service.

III. Impact of the DSM Directive in achieving a well-functioning marketplace for protected content (Articles 15 – 17)

III.1. Article 15 – Protection of press publications concerning online uses

29. To what extent has your national implementation of Article 15 been effective in strengthening the position of press publishers when negotiating with information society service providers (ISSPs) for the online use of their publications? Which implementation issues / challenges have been identified so far in your Member State, if any?

-please consult the response by VG Wort-

III.2. Article 16 – Claims to fair compensation

-please consult the response by DJV-

 

III.3. Article 17 – Use of protected content by online content-sharing service providers (OCSSPs)

[…]

40. Please describe any identified legal or practical issues in applying Article 17 at the national level (e.g., interpretation questions, operational challenges, enforcement/oversight issues). Where available, please refer to concrete sources (guidance, case law, regulator reports) or state “none known to this authority”. Please insert answer here.

-please consult the response by VG Wort-

41. Please list any licensing agreements / frameworks involving OCSSPs that are publicly documented in your Member State (e.g., through published announcements, regulator materials, CMO reports, competition authority decisions, court cases). If none are known or held, please state “none known to this authority”.

– no response available-

III.4. Cross-cutting implementation observations for articles 15 – 17

– no response available-

IV. Impact of the DSM Directive in achieving fair remuneration in exploitation contracts for authors and performers (Articles 18-22)

IV.1. Article 18 – Principle of appropriate and proportionate remuneration […]

COMMENT to the non existing questioning of Article 18 implementation and its impact. 

Although the Directive (EU) 2019/790 states that authors should receive an appropriate and proportionate remuneration, we need to point out that, in Germany, the “proportionate” factor has NOT been implemented. Which leads to a complete dysfunctional contractual rights matrix.

Plus: There is _no_ mutual and clarified definition of what an appropriate remuneration or a proportionate remuneration means. There is also no agreement between writers and publishers on what “appropriate” and “proportional” mean in mere practice.

On the one hand, this dispute in perspective is unsurprising and goes on since a century. On the other hand, it shows that our government in power have little understanding of the ways in which authors live and work. Going to court is no option: an author cannot demand or sue for disproportion between payment and use without being able to relate to “proportionate”.

IV.2. Article 19 – Transparency obligation

[…]

49. Do you have information on how the transparency obligations in Article 19 operate in practice in your Member State? In particular, please indicate: observed impacts, and any practical problems, disputes, or requests for clarification encountered.

RESPONSE to Q49

The implementation in German laws left authors out in the opaque rain of useful transparency and usage documentation. Appropriate remuneration and subsequent renegotiation are based on knowledge of usage figures. Similarly, any collective bargaining rules (CBR) are based on knowledge of general and specific usage figures.

However, in the course of implementing the EU Copyright Directive, the German legislature has made things more difficult rather than easier for creators – in our case, authors and translators.

Summary:

German law makes it possible for a publisher not to provide information or to refuse to do so if it

 (a) is declared ‘unusual’ (who decides this?),

(b) entails additional administrative costs that would exceed the remuneration to be paid (who proves this without any concrete figures?),

(c) a work makes a so-called ‘subordinate’ contribution, e.g. in our cases this may apply to anthologies or authors of forewords – or by the arbitrary decision of a publisher what “subordinate” means.

(d) At the same time, the burden of proof regarding disproportionate and therefore unreasonable remuneration is shifted onto the author or translator; it is up to them to provide evidence. Which is difficult – without information.

A publisher may also refuse to provide information if there was a total buy-out. Or contractually stipulate a sum below which neither information nor payment is possible.

The German implementation gave authors only got a right to transparency, and not the well intended obligation of granting information posed on the exploiter and distributor. Third contractual and distributing parties do not have the mandatory duty to deliver all data, although this is especially needed in digital usage. This also means that the basis of appropriate (and our missing proportionate) remuneration practice is not sufficiently given at all. Without transparency about each use, no appropriate remuneration or even re-negotiation under Article 20 is possible.

When there is a mandatory accounting for authors, only a minority (9.09%) gets two accounting reports per year. Also, only a bit more than a third (36.36%) get third parties’ data (for example on number of loaned titles in commercial flat rates, numbers of streamed audio books, or sales of translations). These numerous national restrictions on the obligation to provide transparent information weaken the important effect that the directive was intended to create in the interests of authors. In general, the lack of transparency of platforms and distribution monopolies is to be criticised. Here, the transparency obligation must also apply to companies outside the EU legal area, and also on state institutions (libraries) – which they do not.

II Current status of availability of information to authors publishing in Germany:

II.1 Print books in the country of publication or the legally designated language area (German-speaking/translations/worldwide)

Information is provided for:

• Total retail sales in Germany, Austria and Switzerland

Information is not provided, or is not specified, for:

• Breakdown by physical bookshops and online retailers
• Sales to libraries (public and university)
• Sales in German-speaking countries abroad (e.g. in Austria or Switzerland, where retail prices are often different)
• Class sets, school textbooks
• Number of batch and postage replacement copies

Information is provided in a heterogeneous / imprecise manner for:

• Sales of translations abroad
• Number of press, review or other promotional copies

Special case: royalties based on net publishing proceeds rather than the net retail price:

• Information is not provided in a specified manner regarding the breakdown of discounts and overhead costs incurred by wholesalers, online retailers, brick-and-mortar bookshops, etc., which are deducted from the net retail price to arrive at the net publishing revenue.

Special case: translations and translators’ share abroad.

• Practices vary in the USA, for example, translators do not receive a share, and if they do, the author is required to contribute 0.5 or 1% of their royalties. In the UK, publishers already allocate an additional 0.5 or 1% (or more) and do not deduct this from the author.

Large publishers settle accounts twice a year; smaller ones once a year. Few have an author portal for real-time digital tracking.

II.2 E-books and further digitised text works

Information is provided for:

• Total sales of reading licences (‘print run’, ‘copies sold’)

Information is not provided specifically for:

• Breakdown by distributor: publisher’s website, portal, online bookshops (and corresponding deductions for calculating net publisher’s revenue, see below)
• Sales of licenses for digital lending in public libraries, including framework conditions, purchase price and license model
• Sales in German-speaking countries abroad (e.g. in Austria or Switzerland, where retail prices are often different)
• Usage data via information on so-called “multiple use” within the context of, for example, “family sharing” and simultaneous reading on up to 5 different devices
• Usage within apps and so-called digest business models (Blinkist, GetAbstract)
• Usage within “AI features”: enhanced interactive e-Books with Chatbots; Recaps; automated summaries; Quotes about the book (Grammarly)
• No information about internal AI uses by the publishers.

Information on e-books is not provided for:

• Discounts, commissions, ancillary costs via digital distributor portals
• Lending transactions in digital lending of public libraries
• Lending or usage transactions in commercial digital flat-rate schemes and shares
• Usage and usage figures for regular text and data mining (not AI use)
• Usage within zero-euro promotional campaigns

II.3 Audiobooks (Stream; Download) and digitised audiobooks on CD

Information is provided for:

• Total sales of purchased (stream/download/physical media) audio licenses

Information is not provided in a specified format for:

• Breakdown by distributor: publisher’s website, portal, online bookshop
• Sales of licenses for digital lending in public libraries
• Sales in German-speaking countries abroad (e.g. in Austria or Switzerland, where net shop prices often differ)

Information is not provided for:

• Discounts, commissions, ancillary costs via distributor portals
• Lending transactions in digital lending of audiobooks in public libraries
• Usage transactions within subscription or flat-rate models (Spotify, Audible) with pooling remuneration schemes (one pot divided through all consumed minutes)
• Framework agreements and revenue models, including usage in shared-revenue models (pool models such as Spotify/Deezer, etc.)
• Framework agreements and usage revenues for providers on a fixed-fee basis (BookBeat, Nextory, Storytell)
• Usage surveys for promotional licenses within commercial enterprises often remunerated on a lump sum, one time payment flat-rate basis (Lufthansa, Deutsche Bahn, etc.), which is a total buy out and harm the principle of appropriate and proportionate remuneration.

Special case:

Reference values for net publisher revenue. In the audiobook sector, the HAP (retail price) is the market standard as a reference value. In the ‘traditional’ print book sector, a retail margin of 40% (calculated on the gross retail price, i.e. including VAT) was often standard; however, this cannot be applied in the digital sphere in the same way, as it depends heavily on the individual platforms or the publishers’ specific deals with those platforms – and these are kept strictly confidential. Thus, in principle, the author or translator has little say in whether they agree to a deal, nor can they assess whether the “appropriate remuneration” applies.

III Desired situation

• All information not yet specified or provided should be disclosed – or a list of mandatory and optional items drawn up as part of joint agreements.
• Likewise, a fundamental agreement must be reached on the specification of the ‘derivative’ portion of a work (which legitimises a refusal to provide information), or, in the best-case scenario, general information should be provided regardless of the scope of the work’s components.
• Third-party license holders:  In the digital sphere, public libraries and commercial (and non-European) audio and e-book portals in particular must be held to account; this is also covered by Section 32e: e-book sales are a significant (co-)determining factor in exploitation.
• In the case of translations, care must be taken to ensure that the right to information of the country of origin is applied.

 

IV.3. Article 20 – Contract adjustment mechanism

52. Do you have information on how the contract adjustment mechanism operates in practice in your Member State? Please indicate any observed impacts, practical problems, or interpretative difficulties that you are aware of.

RESPONSE:

The flawed transparency provision and the lack of having the right of “proportionate remuneration” is hindering authors to get in re-negotiation. In fact, any adjustment of contracts are hindered until today, and often authors have no money to sue, or fear to be put on a blacklist.

Concerning contracts and recommendations, most of the responding authors to our internal monitorings say, they only can rely on recommendations confidentially or individually given, as the still valid competition law do not permit Authors’ associations to give contractual advice publicly.. The existing “norm Contract” (a not so much binding) agreement with publishers (seven altogether from over 10.000 in Germany), sets long overcome standards below an acceptable minimum under the real practices today. Here, the EU-wide modification of competition law on joint remuneration negotiations should close an important legal gap. After all, the CDSM Directive refers to the chances of common rules and collective bargaining seven times. But still the Union law on competition does not provide the legal framework.

IV.4. Article 21 – Alternative dispute resolution procedure

[…]

RESPONSE:

Germany did not install any mediation entity for alternative dispute resolution. But only 90% of the other 26 MS did so. Authors in the book sector are left alone throughout Europe.

 

IV.5. Article 22 – Right of revocation

[…]

55. Do you have information on how the revocation provision under Article 22 operates in practice in your Member State? Please indicate any observed impacts, practical problems, or interpretative difficulties that you are aware of.

RESPONSE: Rights revocation will still remain a question of individual contract negotiations.

Our authors hope, that – often only with the support of tough Literary Agents – an easier recall of rights is contractually laid down, when a single transferred right is not exercised (“use it or lose it” -principle), or a clearer recall of rights when the work is not reprinted for a certain period (for instance, 3 months to 12 months) and the conversion of an exclusively granted right of use into a simple, non-exclusive right of use after a certain time period (for short texts or anthologies; after 2 years).

In practice, unfortunately, the author is often required to send the publisher several written requests, either to reprint the work or to declare that the rights have reverted to the author. In this regard, the German legislature has not succeeded in improving the situation for authors, for example by establishing a clearly defined timeframe that could also have been structured flexibly enough to accommodate license agreements with third parties.

IV.6. Cross-cutting implementation observations for articles 18 – 23

[…]

RESPONSE: Overall, the inadequate implementation of Articles 18, 19 and 22 has led not to an improvement, but to a deterioration in the negotiation and enforcement of authors’ rights.

 

Author: Nina George (Commissioner for Political Affairs)

[1] https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(2025)774095_EN.pdf

[2] https://urheber.info/diskurs/ki-training-ist-urheberrechtsverletzung

[3] https://www.europarl.europa.eu/doceo/document/TA-10-2026-0066_EN.html

[4] https://www.chiplawgroup.com/architect-of-eu-copyright-law-says-ai-loophole-is-irresponsible/

[5] https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(2025)774095_EN.pdf

[6] https://urheber.info/diskurs/ki-training-ist-urheberrechtsverletzung

[7] https://www.europarl.europa.eu/doceo/document/TA-10-2026-0066_EN.html

[8] https://www.chiplawgroup.com/architect-of-eu-copyright-law-says-ai-loophole-is-irresponsible/

[9] ^] https://europeanwriterscouncil.eu/gai-is-based-on-theft/

[10] https://www.vice.com/en/article/v7b774/ai-generated-books-of-nonsense-are-all-over-amazons-bestseller-lists