The Good, the Bad and the Missing – the new proposal for the implementation of the CDSM Directive into Polish law

The Polish government presented a new project for the implementation of the CDSM Directive. We take a closer look at the project: what went well, what was missing again, what is controversial.

Poland has been waiting for the implementation of the CDSM Directive into national law for almost 5 years. During this time some early implementation projects appeared, and in 2022 public consultations were conducted, after which things went quiet. After winning the elections in October 2023, the new ruling coalition embarked on another attempt to regulate these issues. Currently, we are at the end of yet another round of public consultations. In this blogpost we gather the most important highlights regarding the new implementation project.

The “Good” (which could still be better)

In the CJEU judgement in case C-401/19, the Court ruled that, while Article 17 of the CDSM Directive is compatible with EU law, Member States have an obligation to introduce user rights protection both at the stage of content upload on platforms (ex ante protection) and after the content has been uploaded (ex post protection). The action annulment was brought before the CJEU by Poland on the grounds that Article 17 infringed the right to freedom of expression and information, therefore 2, we were surprised by the initial government proposal, which completely ignored the issues of the ex ante protection system, proposing only very limited mechanisms for ex post protection.

The newest implementation proposal finally seeks to address the CJEU ruling. Firstly, it points out that cooperation between online content-sharing service providers (OCSSP) and rightholders cannot lead to the unavailability of works uploaded by users in accordance with the law, especially based on provisions regarding permitted uses of copyrighted works and other subject matter. In fact, it is explicitly stated that “Actions taken by the service provider cannot lead to the automated prevention of access to works, blocking access, or their removal if the public sharing of these works does not manifestly infringe copyright.”

In addition, the online platform is required to promptly inform the user of any instance where access to the work is prevented or blocked, or if the work is removed. While this new provision is important, a more significant step in this direction could have been taken. Perhaps, following the Advocate General’s opinion in the Polish case, it would be worthwhile to explicitly state that blocking must be “limited to content which is ‘identical’ or ‘equivalent’ to works and other protected subject matter identified by rightholders.”1 Centrum Cyfrowe provided a detailed proposal for such a solution in the previous consultations. Moreover, the proposed ex post protection system by the drafters also requires refinement and significantly diverges from the German approach in this matter.

The bad

Unfortunately, the new proposal also includes a number of provisions that are problematic. Let’s examine a few of them.

First, the Polish government proposes that the implementation of provisions regarding Text and Data Mining (TDM) into Polish law shall exclude the training of generative artificial intelligence (GAI) models. This exclusion is intended to apply to both research-oriented TDM and commercial TDM. According to the government, this approach stems from the fact that during the creation of the CDSM Directive, this type of TDM application was not considered, and GAI models were not as prevalent at the time, thus they must be excluded. As correctly pointed out by Paul Keller, this rationale is entirely flawed. Currently, it is widely recognised that the TDM provisions outlined in the CDSM Directive cover the mining of data for the purpose of training artificial intelligence. This interpretation is also supported by the European Artificial Intelligence Act (AI Act). While generative AI introduces complexities in power relations and the distribution of benefits in society and may require a systemic approach, the implementation of TDM into Polish law is certainly not the appropriate avenue for this discussion for several reasons.

To begin with, commercial TDM regulations are based on an opt-out clause. Therefore, if someone does not want their works to be used for the purpose of AI training, they can simply opt out, which is already happening. Excluding this application of TDM, even from the scientific research exception, seems entirely unjustified. The right to research is of constitutional importance and researchers worldwide are conducting studies in the field of artificial intelligence. Depriving Polish researchers of such opportunities not only appears to be unconstitutional but is also directly detrimental to the development of Polish science. If this exclusion was to be adopted, Polish researchers would not only be unable to conduct such research themselves, but the entire international cooperation of researchers from Poland with other countries would be jeopardised. This issue has already caused significant concern in the academic world, and raised demands for the government to quickly withdraw this proposal during the revision process.

Second, the proposal is not compliant with Article 8 of the CDSM Directive. The CDSM Directive introduced a system for the use of out-of-commerce works (OOCW). Importantly, this change was made in response to the practical inapplicability of the directive on orphan works. The adopted system for OOCW was intended to facilitate real access to those works that would otherwise not be accessible because no one is currently interested in bringing them to market. Under this system, cultural heritage institutions can provide access to such works either based on a licence with representative collective management organisations (CMOs) or, in the absence of such CMOs, based on a relevant exception. This dichotomous division is quite clear and allows for easy determination of whether specific works can be made available. Simply put, one only needs to check if there is a representative CMO. However, according to the Polish legislator, cultural heritage institutions will no longer only have to determine whether there is a representative CMO but also whether a given creator is not represented by any CMO. If a creator is represented by any CMO, it would be impossible to use the OOCW exception. In practice, this introduces the necessity of conducting rights clearance for each creator separately and determining whether they are represented by any CMO and to what extent. Ultimately returning to an unworkable exception, as was the case under the orphan works directive.

Third, and finally, the issue of protection against exceptions’ contractual override and their blocking using Technological Protection Measures (TPM) remains untouched as, for no particular reason, the Polish legislator decided not to implement Article 7 of the CSDM Directive directly. Concerning TPMs, according to the drafters, appropriate regulations were already introduced during the implementation of the InfoSoc Directive, and since they apply to the new exceptions, there would be no need to change anything in this regard. Unfortunately, the implementation under the InfoSoc Directive is rudimentary at best and is essentially limited to the issue of liability for circumventing TPM.

When it comes to protecting exceptions against contractual override, according to the legislator, under Polish law, all exceptions have a mandatory character (ius cogens). Indeed, this also applies to those mentioned in Article 7 of the DSM Directive. While this is assumed by the “dominant” part of legal doctrine as well as the Polish legislator (as evidenced by the legislative proposal), we are not aware of any Polish court ruling that would confirm or deny such an approach.

We generally agree that this should be the case, however,  in order to create sufficient legal certainty, this issue should be explicitly settled in the law. Importantly, it’s not just about meeting the requirements set out in the Directives. People and public institutions make use of provisions for fair use. The risk of copyright infringement, which can lead to liability without fault, is significant due to the complexity of the matter and the three-step test, which does not facilitate its application. Assuming that such “users” of the law will base their knowledge of its content and principles on something other than the text of the law is very problematic. After all, in the normal course of action, no one will consider what legal scholars think about this matter or what the legislator wrote in the rationale. For a long time, it has been recognised that creating clear norms regarding fair use is crucial for its practical application. As such, introducing this additional layer of legal uncertainty may well have a chilling effect on the use of works.

The Missing

At this point, it is worth emphasising what is missing in this proposal: Namely, a vision for balancing the rights of rightholders and users fitting for a  regulation in the 21st century. On the one hand, it is understandable that the new government, which must urgently implement a Directive whose implementation obligation passed years ago, does not have much space to think about the entire ecosystem. On the other hand, the ministry that prepared this project should strive to introduce the best possible regulations. For a long time,  the need to search for legal instruments that would strengthen user rights was strongly identified by social organisations, in academic literature and CJEU jurisprudence. In particular, exceptions and limitations to copyright have long been recognised in CJEU case law as rights2. However, to move from this rhetoric of rights to actual rights, concrete regulations at the national law level are needed, in particular where rightholders prevent uses of works that are in accordance with the law. Of course, in this context, it is also worth considering provisions to counter abuses of copyright.

Finally, it would be desirable to introduce regulations enabling e-lending of books by libraries based on books digitised by them. It seems that this is possible under the currently applicable EU law. The case for such a possibility is supported by the results of research conducted by the Centrum Cyfrowe in cooperation with the Jagiellonian University under an international study of digital lending regulations carried out for Knowledge Rights 21. Similarly, it would be worthwhile to use this opportunity to expand the provisions regarding the use of works for the purpose of conducting scientific research, rather than attempting to introduce a quantitative limitation on how much can be used within such activities (20%) – as currently proposed by the legislator. In this context, it would also be worthwhile to introduce the Secondary Publication Right (SPR) into Polish law. SPR are “the rights of the researchers and their employers to reuse a work after the first publication of a formal version; in this case to republish it on an  infrastructure  that  by  its  nature  (public  and  not-for-profit)  is  coherent  with  the (public) nature of the funding source.” 3 Of course, it would also be worth considering the possibility of creating safe harbours for cultural heritage institutions regarding their good faith use of works for the fulfilment of their public mission, or the possibility of internet archiving by these institutions. In general, it is essential to consider an internal balance within copyright law so that the constitutional rights of rights holders are properly balanced with the constitutional rights of users. The new Policy Recommendations of COMMUNIA can serve as a constant inspiration for this.


  1. Opinion of Advocate General Saugmandsgaard ØE delivered on 15 July 2021 (1), Case C‑401/19, point 201
  2. JUDGMENT OF THE COURT (Grand Chamber), 29 July 2019, In Case C‑469/17, point 161.
  3. Tsakonas, G., Zoutsou, K., & Perivolari, M. (2023). Secondary Publishing Rights in Europe: status, challenges & opportunities. Zenodo., p. 18.


Konrad Gliściński, PhD – expert of the Centrum Cyfrowe Foundation, Jagiellonian University, Faculty of Law and Administration, Intellectual Property Law Chair