Tag Archives: text and data mining

Text and Data Mining: (Articles 3 and 4 of the EU-DSM) by REBIUN’s Copyright working group

The Copyright working group of REBIUN (the network of university libraries in Spain) is formed of Silvia Losa, as coordinator of the group, and librarian in the Universitat Pompeu Fabra in Barcelona, Paloma Jarque, librarian in the Universidad Carlos III in Madrid, Rosa Mª Sánchez, librarian in UNED, and Patricia Sanpera, librarian in the Ilustre Colegio de la Abogacía de Barcelona. The group studies topics of interest on copyright for university libraries in Spain. We are currently monitoring the transposition process to guide REBIUN in the actions to be carried out in order to get legislation in line with the interests of libraries.

  1. Can you explain to us what Articles 3-4 of the EU-DSM Directive are?

Articles 3-4 of the DSM Directive introduce two exceptions to copyright for text and data mining.

Text and data mining (TDM) is defined as “any automated analytical technique aimed at analysing text and data in digital form in order to generate information which includes but is not limited to patterns, trends and correlations”.

Article 3 focuses on text and data mining for the purposes of scientific research.

The article covers the reproduction, and extraction from databases, made by research organisations and cultural heritage institutions (and their members) but only for scientific research purposes. It also covers the storage and retention of copies, for the same purposes, including the verification of research results.

A cultural heritage institution includes “publicly accessible library or museum, an archive or a film or audio heritage institution”. Art. 2(3)

Research organisations are basically not-for-profit entities or entities tasked by a Member State with a public service research mission, according to art. 2(1).

The exception covers text and data mining of “works or other subject matter to which they have lawful access”. That means all the collections of institutions like libraries but also those contents freely available online.

This exception is not subject to remuneration (recital no. 17) and is protected against contract override. Art. 7(1)

Rightholders may establish measures to ensure the security of their systems but they should not prevent the application of the exception. Copies generated by text and data mining should be stored securely. Member States may regulate both aspects after negotiation with stakeholders (including, therefore, libraries).


Article 4 allows acts of “reproductions and extractions of lawfully accessible works and other subject matter for the purposes of text and data mining”.

Text and data mining can be done for any purpose and the reproductions “may be retained as long as necessary for the purposes of text and data mining.” Art. 4(2)

The exception benefits all kind of users, institutions or individuals, who have lawful access to contents. That means all the collections of the organisation but also the open web.

This exception, unlike the previous one, can be overridden by contract.

According to art. 4(3) “the exception or limitation shall apply on condition that the use of works […] has not been expressly reserved by their rightholders in an appropriate manner, such as machine-readable means in the case of content made publicly available online”.


  1. Why are these items important to libraries?

An exception for ‘text and data mining’, TDM, as stated in articles 3 and 4 of the EU-DSM Directive, grants libraries the right to mine in copyright works to which they have lawful access.

Text and data mining, TDM, is important for research and academic libraries because this exception allows them to support researchers and other legitimate users from different disciplines to undertake data mining. This support includes giving them access to legally accessed materials, not only on-site but remotely, and with the right to keep secure copies.

There are some aspects of the activity of libraries that can be closely related to text and data mining.

Libraries are supporters of Open Science, as they do with their institutional repositories. Open Science, including, inter alia, open access, open data, and FAIR data, is a loyal friend for TDM. With such a friend, researchers and other legitimate users will successfully carry out automated text and data analysis. Open Science is based on the possibility of checking out researchers’ methods and data. Without the opportunity to look at the datasets used for analysis, other researchers cannot confirm, or disapprove, findings, undermining overall scientific progress.

Libraries are used to work together with IT and Legal Departments. For the sake of an ideal use of the exception in favour of researchers and other legitimate users, libraries can help TDM workflows and infrastructures to be applied and developed.

As beneficiaries of the exception, and as advocates of researchers and other legitimate users from their institutions, libraries can have the necessary power when negotiating with publishers, so the right to mine is not overridden by contracts, and no additional information about the research is requested by publishers. And, as well, ensuring that any technical issues or access-blocking experienced by the institution are resolved quickly. Libraries pay for subscriptions to academic publications, there is no need to pay again to text and data mine contents already subscribed.

Furthermore, with a TDM exception libraries could, in short terms:

–       Perform TDM without requirement to inform or seek permission from publishers

–       Remove or ignore contractual provisions in licenses in conflict with TDM

–       Promote actions (including legal action) if access is blocked and not quickly resolved by the publisher

–       Protect personal data and privacy of researchers and other legitimate users from publisher requests for further information about TDM activities


  1. What is the best implementation you could hope for with these articles?

In short, our aspiration would be that the legal text allows the maximum use of text and data mining techniques for research purposes, and also to the legitimate users; with the only limitation that such uses do not conflict with a normal exploitation of the works and do not unreasonably prejudice the legitimate interests of the rightholders.

Specifically, we believe that there are a number of issues that it is important to incorporate or clarify in the law:

Data mining exception should allow acts affecting the right of transformation. It is not always clear when the use of these techniques can affect this right, so the express inclusion of this right would create legal certainty for researchers and legitimate users.

Public communication should also be allowed to enable researchers to carry out text and data mining activities where they have better tools for this, through a remote controlled system. That would prevent them from having to move, for example, to library facilities in order to analyse digitisations of their collections.

It should also ensure that the application of the exception entails the possibility of disseminating the results generated by it provided that such dissemination does not conflict with a normal exploitation of the work and does not unreasonably prejudice the legitimate interests of the author.

With regard to libraries and other cultural heritage institutions, the law should specify that they may use the exception of article 3 to conduct research in the context of their main activities. A restrictive interpretation of the concept of scientific research will make the exception useless for our institutions.

The storage of copies generated by the mining of text and data should be made where the researcher or the legitimate users choose, provided that they are protected against unauthorized access. Moreover, imposing different storage conditions in each European country may be an impediment to the development of cross-border projects.

It must be ensured also that, in the case of technological protection measures, the beneficiaries of the exception may have an extraordinary remedy requiring rightholders, or their intermediaries, to lift such mechanisms within a maximum of 72 hours, including financial penalties in case of non-compliance, where appropriate.

Since the exception (both exceptions of article 3 and 4) should not be subject to fair compensation, it should be ensured that suppliers of works and services do not impose a higher price on their subscription to enable text and data mining activities.

Finally, regarding specifically article 4, and according to the EU-DSM Directive, the law should also ensure that in cases of accessible resources that have been made publicly available online, rightholders can only object to the exception through the use of machine-readable means; otherwise the exception will become useless, as a manual review of terms of use and legal notices of websites cannot be intended.

  1. What is your government’s position on the issue?

We have no information about this aspect at the moment. The government launched a public consultation on December 2019 but they did not expose any kind of explanation or clarification on the positions of the government regarding the transposition of the EU-DSM. As far as we know (https://www.notion.so/Spain-64ff430a3fec4ed2a17895bd82ceb6e8), they will probably publish a draft of the legislative text when the State of Alarm ends.


Catching up on Copyright: Current Global Trends in Legal Reform

As will have been highlighted by many of the posts made today – World Copyright Day (23 April), copyright has had a major influence on the way the modern creative industries have emerged.

It has spread way beyond its beginnings in the early 18th century as a limited protection lasting just 14 years, and now reaches into almost every corner of our lives.

It is so powerful, indeed, that even from the beginning of international law-making in the area, it was recognised that it should not be absolute. Just as it enabled authors and creators – or often the companies that bought and traded their rights – to earn a living, it could also bring dangers.

Copyright legislators have therefore sought to intervene in order to help the system work most efficiently. This comes both in terms of finding ways to simplify the way the system works where it is appropriate to charge for access to – or use of – works, and to create and enforce exceptions there would otherwise be market failure.

So how have things been developing since the last World Copyright Day? Below we set out a few trends:


Sorting out the Basics: despite these being seemingly core elements of free speech, it’s not in every country that there is the possibility freely to quote someone else, or to carry out criticism, make jokes, or review. The lack of such rules can allow copyright too easily to become a tool of censorship and control.

Fortunately, there have been steps forwards in Kenya and Myanmar (which previously was working with a law from 1911 inherited from the colonial period). In both countries, exceptions allowing for quotation, parody and criticism have been put in place.

However, some countries still lag behind, notably in Latin America where a number of states still have no meaningful copyright law containing exceptions. Long efforts by the library sector in Uruguay have been frustrated for now, with the government simply pushing through an unnecessary term extension, while it remains unclear what the Brazilian government will propose. Argentina too, beyond a welcome move to allow enjoyment of the rights created by the Marrakesh Treaty, still has some way to go before a full and modern copyright law is in place.


Enabling Digital Uses: a core focus of much library advocacy around copyright reform has been the drive to update copyright laws for the digital age. Provisions in existing laws – limiting the number of copies taken, specifying a format or method to be used, or explicit exclusion – can all make it impossible for libraries to take advantage of new opportunities.

There has been progress over the past year. Text-and-data mining (TDM) has proved to be a continuing area of uncertainty for many, given that while copies are made, these are usually exclusively part of the analysis process.

To resolve this, we have seen Switzerland and Ireland introduce or strengthen laws which make it clear that this is permissible, at least for non-commercial purposes. Singapore has gone further still in its own proposals, underlining that it should be possible regardless of the purpose, given that TDM causes no harm to original markets, and efforts to control it would likely limit innovation. Myanmar’s new law also opens the way to TDM.

Canada’s Industry, Science and Technology Committee also called for a broad TDM exception in its review of the country’s 2012 reforms. As focus increases on artificial intelligence – which often draws on TDM in order to train machines – it can be hoped that other governments will understand the need for strong exceptions in order to facilitate innovation in this space.

Similarly, there have been steps towards facilitating digital preservation. Myanmar, Switzerland and Ireland have both expanded exceptions to allow for preservation copying through digitisation, giving them more scope to safeguard their history for the future. Draft laws in the Philippines look to do the same.

However, not all is positive, with Kenya missing an opportunity to expand its own preservation provisions. Sadly, this will help continue the imbalance in laws that will give institutions in some countries a greater ability than others to fulfil their mission as guardians of memory.


Closing Loopholes: a major concern linked to greater reliance on digital resources is the opportunity that technological protection measures (TPMs), and the primacy of freedom of contract provide to hollow out exceptions.

TPMs can create practical restrictions on how libraries or their users make use of works, even preventing activities that would ordinarily be protected. Badly drafted laws will nonetheless criminalise the circumvention or removal of such measures.

Meanwhile, unless governments make it clear that they are not enforceable, the terms of contracts for digital content can be used to forbid uses, such as lending or document supply.

We have seen more progress on TPMs in the last year. In both Kenya and Myanmar, it has been made clear that libraries and others need to be able to enjoy copyright exceptions. The proposed Lebanese law includes the same idea, and we have seen calls for this in Canada.

Nonetheless, this is not the case everywhere. The Philippines draft law only refers to penalties for circumventing TPMs, rather than offering clarity to libraries. We will also need to see how European countries, in implementing the new copyright Directive, look to ensure that TPMs cannot be used to frustrate legitimate activities.

Meanwhile, the very promising provisions in South African law, ensuring that rightholders cannot use their negotiating power to deprive libraries and users of their rights under copyright exceptions, are still in limbo due to delay by the President in signing the law.

Unfortunately, neither Kenya nor Myanmar included provisions to prevent override by contract. However, we will see a number of European countries obliged to introduce such measures as they implement the Directive on Copyright in the Digital Single Market. We can hope that they will take the opportunity to extend this to other exceptions enabling public interest activities.


Realism on Rights Management: The argument that collective management can remove all of the complexities involved in making copyright work can be attractive. With many small (and a few large) producers trying to engage with many small (and a few large) consumers, an intermediary can make life much simpler. We have seen collective management organisations (CMOs) given new powers to offer licences in Kenya and Switzerland for example.

However, this is also a role of major responsibility. Good collective management can help realise the promise of copyright, and ensure that creators are fairly remunerated. Bad, over-reaching collective management can serve to bully libraries and users, undermine the public interest goals of exceptions, and still do little to improve the lives of creators.

Many governments are coming to see the need for tighter regulation and control of CMOs, with new rules brought in in Kenya and Singapore, and proposed in the Philippines. In both Singapore and Canada, as well as in Australia, there has also been a readiness to dismiss arguments made by CMOs when not based in fact.

Governments will need to act as strong and independent regulators of CMOs in order to ensure that they fulfil their positive potential and do not end up undermining the public interest side of the copyright equation.


Conclusion: 2020 and Beyond: the examples shared so far all date from pre-COVID-19. Understandably, since the pandemic took hold, the primary area of focus has been on providing healthcare and dealing with immediate human challenges.

However, the COVID-19 pandemic is clearly asking some serious questions of copyright systems. In many cases, rightholders have indeed been willing to find ways to give access when this would otherwise have been impossible when library doors have closed. This is very welcome.

However, arguably, being able to access something that has already been paid for, or to carry out an activity which would be permitted without question in person, should not rely on the goodwill of private actors.

This reliance – and the variety of responses made – has added to the disruption already caused to education, research, and access to culture.

The failure of copyright laws in general to adapt to the digital age, ensuring that libraries and their users do not have to play by different rules when using different formats, has been made very clear.

World Copyright Day 2020, we can hope, will be a wake-up call for law makers.

At the Heart of the Response: Health Librarians Support Better Decision-Making around COVID-19

For library and information workers around the world, the main challenge faced is how to continue providing usual services in extraordinary times. In order to minimise disruption to education, research and access to culture, great efforts are being made to address legal, financial, technical and practical challenges.

Yet for some in the library field, these extraordinary times have also brought extraordinary demands and pressures. Health librarians – working in hospitals, research centres and governments – are having to deliver more than ever, even as they face the same restrictions and rules as everyone else.

With today – 7 April – being World Health Day, it is therefore a good opportunity to look at and celebrate their work.


Supporting the Decisions that Matter

It was the coming into force of the Constitution of the World Health Organization (WHO) on this day in 1948 that provides the justification for 7 April being World Health Day.

The WHO itself has a strong focus on the importance of information in effective health policies, as well as a very active library which not only acts as a hub for knowledge, but also for its dissemination and application through partnerships and networks globally. This work has been essential as eyes turn to the WHO’s own website, and the advice given there, all based on scientific literature gathered by the team there.

Elsewhere, librarians at the National Library of Medicine in the United States – in particular through the PubMed Central platform, have been supporting vital access to evidence for decision-makers. They have also worked closely with publishers in order to make articles and collections open that would otherwise have been paywalled.

Crucially, they have also worked to underline that collections need to be available in machine-readable format through the COVID-19 Open Research Database. This is essential if researchers are to be able to carry out text and data mining in order to identify potential treatments or cures.

Furthermore, the Library has also acknowledged the importance of discoverability, highlighting tools available for identifying relevant sources on its website.

Clearly, a key contribution to the discovery and application of information comes from it being presented in ways that work for those who need to use it. Here too, health librarians are playing a key role.

From daily briefings to both government and medical decision-makers to more in-depth reviews of the literature on emerging issues, librarians are helping to inform choices made. For example, Public Health England’s Knowledge and Library Services team is producing regular reviews of emerging evidence, while the Irish National Health Library and Knowledge Service is sharing rapid evidence reviews, and the Health Libraries Group of the Australian Library and Information Association has compiled live responses to key literature searches. In Iran, librarians are also supporting efforts to make sense of the existing literature around coronaviruses.


An Informed Public

One of the key lessons already from the response to the COVID-19 Pandemic has been the importance of the actions of individuals. With health systems struggling with the rapid spread of the virus, it has been clear that people need to change their habits and behaviours, distancing themselves from others.

Public health ministries and agencies – again with the help of librarians – have been working hard to produce clear and meaningful information for the public, explaining the situation and the responses needed. This has, for example, been helpful for the library field in understanding the risk of contagion via surfaces such as books or computer mice.

There is also a role – not just for health librarians, but for the library field as a whole – in promoting wider health literacy. When people can understand the global situation, and how and why they should act themselves, the job of those in charge of ending the pandemic is clearly easier.

Of course – just as in the case of decision-makers – the spread of this information and these skills depends often on how well adapted they are to the target audience. Simply placing things on a website may not be enough, especially for users who may have limited digital skills or even no access to a computer.

This is of course another area where libraries have a unique role to play as community organisations. It is also the subject of a webinar organised by IFLA’s Evidence for Global and Disaster Health Special Interest Group and Health and Bioscience Libraries Section, due to take place on 23 April. This will look at the lessons that can be learnt from past practice, and what more libraries can do to make sure that all members of society have the information they need to cope in these difficult times.

Join us, find out more, and share your ideas on 23 April!

7 out of 10: the ARIPO Model Copyright Law

The African Regional Intellectual Property Organisation (ARIPO) has released its model copyright law.

ARIPO it aims to support the work of intellectual property (IP) teams across Africa, through both country-specific capacity building, and regional-level reports and guidance.

Its 19 Member States come primarily from English-speaking Africa (with some exceptions), and will now doubtless be encouraged to refer to the Model Law in reflecting on their own reforms.

This means that the document has a potentially powerful impact. As such, it is worth being clear about its strengths, weaknesses, and silences, from a library point of view. Library associations and others advocating for better laws for libraries should be aware of where the Model Law will, or will not help.

This blog therefore explores the positives, the negatives, and the holes in the Model Law. All references to Articles are to the Model Law, unless stated otherwise).


The Good

Fair Dealing: in the first Article of the chapter on exceptions and limitations (Article 18), the Model Law suggests that uses which constitute fair dealing, for the purposes of scientific research, private use, criticism or review, or the reporting of current events should be permissible (Article 18(1)). It then offers a set of criteria for judging the fairness of this dealing – the purpose and character of the use, the nature of the protected work (how original is it?), the amount of the work used, and the effect of the use on the market (Article 18(3)).

This is a positive step, giving valuable flexibility to libraries and other users in making reasonable uses of works. However, it if course remains less open than fair use exceptions, which do not have closed list of accepted purposes. If the list was to be made open (for example by adding a ‘such as’ before the list of purposes), it would offer an even better model.

Inclusion of Unpublished Works: too often, copyright exceptions only apply to works which have been formally published. This can make the work of libraries and archives in dealing with unpublished works more complicated. The Model Law underlines that the fact of being unpublished does not prevent uses under fair dealing (Article 18(4)).

A Digitally-Reading Education Exception: the Model Law makes it clear that it is possible to make uses of copyrighted works for education purposes via electronic networks, and not just in analogue form (Article 21(1)(a) and 21(1)(b)).

Interestingly, the only area where the exception can be cancelled out by a licence is for in-person teaching (Article 21(1)(c)(iii)). This is clearly not ideal, given that the Article is, anyway, covered by the rule that uses under exceptions should not conflict with the normal market exploitation of a work.

A Technologically Neutral Definition of Copying: too often, national laws suggest that copies can only be made through a specific technology, such as photocopying. The Model Law has the merit of underlining that it is possible to make reproductions through any format (Article 2).

Protection of the Public Domain: the Model Law includes standard provisions on facts, data, news and political speeches not being protected by copyright, but is clear that this also applies to laws, court judgements and other administrative texts are also in the public (Article 6). Furthermore, there is an explicit definition of the public domain, which allows the possibility for authors to renounce their rights. This is positive, given the tendency in some countries to create unwaivable rights which undermine initiatives such as Creative Commons (Article 35).

No Term Extension: the Model Law does not take the opportunity too often used elsewhere (and in spite of the evidence) to go beyond protection lasting for the lifetime of the author plus fifty years. This is a useful model to use elsewhere.

Inclusion of Museums: the provisions on library copying also apply to archives and museums. This is a positive, given the challenges identified in WIPO work around museums facing different conditions and rules to other heritage institutions.


The Bad

Overall, the Model Law provides a relatively good example for governments. However, there are some weaknesses which libraries should look to avoid replicating in their own national legislation. The below suggestions are in addition to the encouragement to adopt fair use above.

Vague Provisions on Circumventing Technological Protection Measures: in line with the WIPO Copyright Treaty, the Model Law underlines that ‘effective’ technological protection measures should themselves be protected by law. In other countries, there then follows a guarantee that users should be still be allowed to carry out permitted acts (i.e. under exceptions). However, the ARIPO Model Law only provides that governments may make exceptions. This is far too weak at a time that libraries are acquiring a growing share of collections in digital form (Article 40(4) ad 45(3)).

No Lending Exception: the Model Law includes public lending as one of the uses over which a rightholder should have exclusive rights. This is not something required by the Berne Convention itself (which only covers rental). This risks obliging libraries to make payments or seek authorisation for lending (over and above what they have paid to acquire books in the first place). This risks seriously damaging libraries’ ability to promote literacy and a love of books (Article 7(1)(k).

Restrictions on Preservation Copying: while the Model Law does (commendably) not limit the technology used to make copies, the fact that it only talks about ‘a’ copy poses to digitisation efforts (Article 23(3)). In line with recent EU reforms, it would be better to talk about taking copies in the quantity necessary to achieve the goal.

Furthermore, the Model Law also includes the obligation to see if a commercially available copy is available before taking such a copy. This risks introducing an unhelpful administrative burden, and may not be practicable. Given that it is usually cheaper to buy a copy than digitise and preserve, it would be better to leave the choice between copying and buying to libraries, rather than enforcing it through law (also Article 23(3)).

Imposing Commercial Availability Checks for Marrakesh Copying: The Marrakesh Treaty made an important breakthrough by removing copyright-related barriers to making and sharing accessible format copies of books for people with print disabilities. It did however leave the possibility for Member States to impose restrictions though, in the form of an optional remuneration requirement, or the obligation to check if an accessible format copy is not already available on the market before making or sharing one (Marrakesh Treaty, Articles 4(4) and 4(5)).

Libraries have argued strongly against making use of either of these possibilities, given the financial and administrative cost. However, the Model Law does suggest that there should be a check on commercial availability. Given the lack of information about which books are available where, and in what formats, such a requirement risks only leading to uncertainty.

Lack of Provisions on Collective Management: the Model Law is surprisingly thin on guidance about the regulation of collective management organisations (Article 57), while at the same time including provisions on extended collective licencing (Article 38). While it is clear that well-managed collecting societies can facilitate the work of libraries when carrying out uses that fall outside of exceptions, it is essential that these are run in a transparent and accountable way in order to be legitimate.

The Model Law says very little about the need for CMOs to be independent of government (in order to avoid conflicts of interest in the operation of copyright offices), to publish information about how much they are collecting and paying out, or to be representative of rightholders and rights when offering licences. At a time when multiple governments are needing to act to force better governance in this field, the vagueness of the Model Law is troubling.

Over-Application of the Three-Step Text: The Berne Convention only applies the three-step test (that a use needs to be a certain specific case, not conflict with the normal market exploitation of a work, and not unreasonably prejudice the legitimate rights of rightholders) only to exceptions to the reproduction right (Article 9(2), Berne Convention). However, the Model Law applies this to all exceptions, leading to potentially unhelpful debates about what is and is not possible (Article 18(2))

Attribution Obligations: on various occasions (Articles 18(1), 21(2), 22, 23(2), 24(1), 26(3)), the Model Law suggests that use can only be made under exceptions if there is attribution. However, this may not always be possible. Laws elsewhere recognise this possibility to make uses without attribution when this is not practicable. However, the Model Law does not, creating uncertainty for users who do not know the author of the work they are using.

Licence Override for Document Supply: as mentioned above, the exception allowing for educational uses of works in face-to-face teaching can be disapplied when a licence is available. This also applies to situations when libraries are making copies for the private use of users. Where a collective management organisation argues that it can offer licences, this could do a lot of damage to document supply activities (Article 23(2)(a)(iii).

Block on Parallel Importation: in the context of WIPO, libraries have argued that even when there is a domestic rightholder with the right to distribute a work nationally, it should be possible for libraries to make acquisitions across borders. This can be essential in order to meet requests for specific versions of works, or, for example, when the domestic rightholder is not active. The Model Law gives the rightholder the exclusive right to import works, without exceptions (Article 7(1)(j)).

Limits on Caricature, Pastiche and Parody: the Model Law does include welcome exceptions to economic rights (such as reproduction) for review and critic. However, the exception for caricature, pastiche and parody (Article 30) only applies to moral rights, and not to economic rights. As such, it may make it possible to restrict such activities on other grounds.


The Missing

The Model Law, while comprehensive, does not cover a number of areas which, from a library point of view, would be desirable in any national copyright law.

Orphan and Out-of-Commerce Works: libraries hold many works which are no longer commercially available, but are still in copyright. As such, there are significant restrictions on how far they can give access. This is particularly difficult when a work is orphaned (i.e. it isn’t possible to identify or local rightholders). More and more countries are introducing provisions allowing libraries to permit use of such works, subject to various conditions. The Model Law does not even reference these issues.

Text and Data Mining: legal uncertainty about the possibility for libraries to allow for text and data mining of works in their collections has lead a number of countries to introduce explicit exceptions. There is nothing about this in the ARIPO Model Law, meaning that there is continued uncertainty.

Limited Exceptions for People with Disabilities: while the Model Law does copy provisions from the Marrakesh Treaty for people with print disabilities, it does not take the opportunity (foreseen in the Marrakesh Treaty) to apply similar rules for people with other disabilities (such as sub-titling for people experiencing deafness). Many countries do allow copying without restrictions on the type of disability – it is a shame that this possibility has not been included in the Model Law.

Contract Override: the Model Law is silent on the issue of contract override – i.e. the possibility for exceptions and limitations to be cancelled out by the terms of a licence. In a growing number of countries, there are conscious steps to prevent this from happening, and so defend user rights. National governments should introduce broad provisions ensuring the pre-emption of any contract terms which do undermine exceptions.

Cross-Border: the Model Law only refers to cross-border uses in the case of the provisions on sharing accessible format copies of works for people with print disabilities. There is nothing anywhere else which would allow for the cross-border application of exceptions.

This is perhaps an unfair criticism of course – it is only through international law making that there can be legal certainty for cross-border uses. ARIPO itself could act, but the most effective solution would need to come from WIPO itself. IFLA of course continues to engage to achieve this.



Overall, the ARIPO Model Law does cover a number of key points which help libraries do their job, in particular relatively flexible fair dealing provisions. However, there remain a number of flaws, both specifically (lending rights, limitations on preservation), and cross-cutting ones (contract override).

Governments should therefore not look to adopt the Model Law wholesale, but rather work with their library associations to ensure that they have rules that truly support the public interest missions of libraries. Overall, the Model Laws gets a 7 out of 10.

The EU Copyright Reform: Battles Won, Bullets Dodged, and the Questions that Remain

The text adopted yesterday at the Council, after the Parliament’s vote on 26 March, comes after years of discussion.

The Commission’s proposal released in 2016 has been reviewed and voted upon in five parliamentary committees, at the plenary of the European Parliament, at several levels within the Council, and through trilogue discussions among the Council, the Parliament and the Commission.

IFLA and its partners have engaged in every step of the process to ensure the best results possible for our sector. There are also a number of very good analyses out there, not least those produced by our partner organisations.

This blog, rather than looking to give a comprehensive overview of the legislation, looks at the battles that libraries have, together, won, the bullets we have dodged, and the questions that remain. These last will be important, with the effort to ensure the right rules for libraries now moving to the national level. You can access the final agreed text here.



First of all, the battles won – those areas where we have seen a significant improvement on the text originally proposed by the Commission in September 2016. Through the work of a number of committed Members of the European Parliament, and supportive Member States, there have been important steps forward in some key areas:

  1. Extension of the mandatory text and data mining (TDM) exception to all libraries and cultural heritage institutions (Article 3): originally, the Commission proposed that only research organisations could benefit. However, research libraries may have faced a lack of certainty, and other libraries and cultural heritage institutions would have been excluded, forced to seek licences in order to carry out analysis on the works they hold. Now they can.
  2. A second and broader mandatory TDM exception, applicable to any individual or institution (Article 4): the first version of the TDM provision created extensive uncertainty by trying to distinguish between types of mining which would and wouldn’t be allowed. The final version of the Directive makes TDM exceptions mandatory in all Member States, for all users with legal access, albeit with some smaller limitations. See below for more.
  3. Clarity on cross-border networks of collaboration for preservation (Recital 28): while the need for cultural heritage institutions to work across borders in order to make the most of digitisation equipment was mentioned in the original impact assessment, the first version of the Directive failed to make it clear that such networks were possible. Now it does.
  4. Application of contract override to preservation (Article 7(1)): libraries can be prevented from carrying out preservation activities when the terms of the licences under which works are accessed state otherwise. The original Directive did not address this issue. Now it does.
  5. A wider number of purposes acceptable under digitisation for preservation (Article 6, Recital 27): preservation does not only imply taking a copy of a particularly vulnerable work. Cataloguing, insuring and even lending to another institution for preservation work may also require copies to be made. Improvements to the Directive have offered a greater indication that these are possible under an exception, although libraries will need to work at the national level to ensure these are covered.
  6. A fall-back exception for out-of-commerce works where no collective management organisation exists for a specific category of work in a given country (Article 8(2)): the grand plan in the original Directive was to allow for extended collective licensing of out-of-commerce works? But what about the many sectors and countries where there isn’t a representative, well-governed collecting society to run this, or they don’t have the right mandates? Thanks to the new exception, libraries can now also find a way to digitise and make available works which aren’t available on the market.
  7. Stronger conditions on when a country can opt out of the education exception (Article 5(2)): the original proposal left a lot of scope for Member States to disapply the new education exception and allow licensing to prevail. However, it is clear that many educational licences are not fit for purpose. The final version of the Directive puts the onus on Member States to ensure that before an exception is taken away, licences have to offer a realistic solution.
  8. Protection of the public domain (Article 14): recent cases have seen actors take simple photos of works which have long been in the public domain and claim copyright. This can represent a barrier to their spread, as key texts and images risk being subject to infringement proceedings. The final version of the Directive makes it clear that straightforward reproductions of works in the public domain cannot themselves claim copyright.
  9. A clear possibility to have broader limitations and exceptions (Article 25): the tendency in international copyright law is to favour higher levels of protection of rights, rather than greater scope to pursue public interest goals through exceptions. However, in the final version of the Directive, it is made clear that Member States should feel able to go further if they want.
  10. Extension of education exception to uses by educators in other settings (Article 5(1)): the original version of the directive allowed for teachers to use digital works in the classroom, or online. This potentially restricts the ability of educators to offer courses in libraries and elsewhere. The final version of the Directive clarifies that this is possible under the exception (or licences if applicable).



In the course of the discussions, a number of ideas emerged which would have seriously limited the effect of the new rules, and indeed have created dangerous precedents both for Europe and the rest of the world. Fortunately, they didn’t stick:

  1. Obligation to delete datasets created for text and data mining (Article 3): a number of MEPs tried to argue that if copies of articles and other materials were made in machine-readable formats for text and data mining, these needed to be destroyed afterwards in order to prevent against misuse. This (once again) makes the lazy assumption that exceptions are more or less the same thing as piracy (wrong), and would have meant that experiments carried out with TDM could not be reproduced.
  2. Scientific publications in the scope of the new rights for online press publications (Article 2(4)): in one European Parliament committee, there was an to extend the new planned press publishers right extended to scientific publications. This did not make sense, given the very different market conditions there (not least the fact that authors are not paid for their work). Fortunately, MEPs saw sense and rejected this proposal.
  3. Continued over-protection of technological protection measures (Article 7(2)): the original Directive took the refreshing step of arguing that technological protection measures (TPMs) which prevent the enjoyment of exceptions (for example copying for preservation) should not themselves enjoy legal protection, even for licenced (as opposed to purchased) works. The European Parliament tried to reverse this, leaving any work accessed under licence potentially tied up in TPMs. This proposal did not make it to the final version.
  4. No possibility to cumulate exceptions and limitations (Article 7): a further effort sought to overturn the (highly restricted) TU Darmstadt ruling. This established that it is possible to combine exceptions, as long as these continue to respect the three-step test. Despite this obvious safeguard, rightholders tried to add in a new clause that would prevent ‘stacking’, but which would have at the same time had a huge impact on disciplines such as digital humanities. It didn’t make it.
  5. Automatic right for publishers to benefit from public lending right at the expense of authors (Article 16(2)): one often-overlooked article in the Directive served to protect collective management organisations who had been paying out shares of copying revenues to publishers. Following the Reprobel case in the Court of Justice of the European Union, they risked having to pay this money out to authors instead – the Directive therefore underlines that publishers can claim a share. There were efforts during the negotiation of the Directive to extend this to public lending right, which would have seen authors in a number of countries lose revenue to publishers. The final version leaves the choice to Member States.
  6. Library repositories being covered by new rules on platform liability (Article 2(6)): the first version of Article 13 (now 17) would have meant that any site hosting large volumes of user uploaded content would need to implement filters to check for infringement, or face liability. This would have placed a huge burden on scientific and open education repositories, which play a vital role in giving access to materials. Thanks to extensive work, there is now a clear exception for these, alongside sites such as Wikipedia.
  7. Libraries and individuals being obliged to pay for uses of short snippets of press publications (Article 15(1)): while clearly aimed at GoogleNews, the original version of the Directive gave very broad application to the Press Publishers Right, with non-commercial users such as libraries potentially liable to pay. This would have potentially had a major impact on research work done by librarians for users, as well as catalogues and libguides. Fortunately, the new Directive is clear that non-commercial users are not affected.
  8. Works can only be declared out of commerce when all versions, manifestations and translations are no longer on sale (Article 8(5)): the original version of the Directive indicated that a work could only come under the new provisions when all versions, manifestations and translations were no longer on sale. This would have seriously limited the impact of the Directive, given that different language versions are not necessarily interchangeable, and that researchers may well need a specific edition, and so cannot complete their work with a substitute that is still on sale.
  9. 20-Year duration for press publishers’ rights (Article 15(4)): the first version of the Directive gave press publishers a right for 20 years, despite any evidence of this being proportionate or justified. This would have seriously limited the work of libraries working with the press, as well as research into recent history. In the end, the duration of the right was limited to two years.
  10. Retroactive effect of the new press publishers’ right (Article 15(4): at first, a lack of clarity in the text could have implied that even existing publications would benefit from the new possibilities. This would have put into question work already done using this material, bringing major new uncertainty. The final version of the Directive is clear that there is no retroactive effect.



As is almost always the case with European legislation, a lot comes down to transposition – the steps taken to turn EU rules into national ones. An added factor in this is the lack of precision in many parts of the Directive. This is inevitable, and offers opportunities to ensure positive outcomes. At the same time, it also means that libraries will need to play close attention.

Here are ten more things to watch out for:

  1. Rules around permissible security measures in text and data mining (Article 3(3)): the Directive underlines that rightholders are allowed to take security measures in order to protect their works. At the same time, this should not lead to the cancelling out of the TDM exception in the first place. Finding the right balance here – and preventing overly restrictive approaches – will be important if the exception is to have its full effect.
  2. Rules around opting out of text and data mining for all individuals and institutions (Article 4(3)): as highlighted above, a major step forwards was the mandatory TDM exception for the benefit of people and organisations outside of research centres, libraries and cultural heritage institutions (if they have legal access to the works mined). There is a catch here, in that rightholders can explicitly state that they do not want their works mined. It will be important to work with Member States to ensure that the rules around this are specific enough to mean that opting out is the exception, not the rule.
  3. Definition of who can benefit from the education exception (Recital 20): despite our efforts, the education exception formally still only applies to educational establishments, although it can be used (under the authority of a formal education institution) in libraries and cultural heritage institutions. There is, nonetheless, a possibility to ensure that libraries (or groups that offer training and support to people in libraries) are recognised as educational establishments in national law. This would open up useful new possibilities for libraries to fulfil their potential as places for learning.
  4. Application of the opt-out from the education exception (Article 5(2)): as highlighted above, there is the possibility for Member States to decide that the new education exception does not apply in situations where there are licences adapted for educational uses on offer. There is likely to be extensive rightholder lobbying in favour of excluding broad categories of works from the exception. It will be up to libraries and educators to ensure that the conditions laid down by the Directive (that licences are ‘suitable’, ‘cover the needs and specificities of educational establishments’ and are ‘easily available’) are fulfilled.
  5. Application of the exceptions in the out of commerce works exception (Article 8(3)): a major area for work will be how to define where cultural heritage institutions need to ask for licences, and where they can benefit from the exception in order to digitise and make available out-of-commerce works. It will be up to member states to decide what it means for a collective management organisation to be ‘sufficiently representative of rightholders in the relevant type of works or other subject matter and of the rights that are the subject of the licence’.
  6. Application of rules to out-of-commerce works by third country nationals (Article 8(7)): one weakness of the Directive is the focus on trying to ensure that works by people from outside the European Union are not covered by new rules on out of commerce works. For many European countries, this will be difficult, given that they use major world languages, and so telling the difference between a French and a Quebecois work, for example, may be difficult. Member States will need to take a sensible approach to this point.
  7. Application of the definition of a good faith search on whether a work is in commerce (Article 8(5)): the Directive suggests that before a work can be declared out-of-commerce, a ‘reasonable effort’ must be made to ensure that it is not available to the public through normal commercial channels. Given the difficulties already encountered around the Orphan Works Directive, it will be important to ensure that national implementation does not create disproportionate obligations on cultural heritage institutions.
  8. Definition of which works can be preserved under the preservation exception and of the activities and purposes covered (Article 6, Recital 27): the Directive’s preservation exception, as set out above, does allow for copying for preservation purposes, while leaving the possibility for Member States to pass other exceptions and limitations for internal uses. There is, therefore, a key opportunity to ask for exceptions that allow for any core library uses of works to be covered.
  9. Management of the dialogues planned for the text and data mining exception, the out of commerce works provisions and the upload filters provision (Articles 3(4), 11, 17(10)): a number of articles provide for dialogues between stakeholders on how the rules should be applied. It will be necessary to pay attention to the composition, terms of reference and other aspects of these discussions in order to ensure that the results reflect the interest of libraries and their users. This will, in particular, be the case around protection of freedom of expression under the provisions on platform liability.
  10. Protecting the quotation exception (Article 15): The press publishers’ right creates a worrying precedent for protection being given to ‘short extracts of a work’. This risks affecting how legislators and courts think of the concept of quotation in general, as well as criticism and review. While the scope of the Article in the Directive is narrow, the precedent is certainly worrying.

For more, see our resources page about the European reforms. You can also see analysis and reaction by our partner organisations:

LIBER: New European Copyright Directive: A Detailed Look

EBLIDA: Long Read : Final stretch for the Digital Single Market Directive

SPARC Europe: A new Copyright Legislation for Europe. How will this impact Open Access?

EUA: EU Copyright Directive: EUA cautious about adopted agreement

The Good, The Bad and (Avoiding) the Ugly: A Way Forwards on the Copyright Directive

Discussions around the European Union’s draft Directive on copyright in the Digital Single Market are as tense as ever. Strong divisions have emerged between and among Member States and Members of the European Parliament around controversial proposals for a new press publishers’ right (Article 11) and an (effective) obligation on internet platforms to filter content (Article 13).

These disagreements stand in contrast to the consensus that has emerged around other provisions in the Directive, which will help libraries and cultural heritage institutions in their work to promote innovation, support education and enable preservation and access to heritage.

Such measures, in line with the EU’s own international obligations, cause no unreasonable prejudice to rightholders, and indeed will support creativity and discovery.

The fear must be that a failure to find agreement on Articles 11 and 13 will lead to calls for the rejection of the Directive as a whole. This would be a huge loss for innovation, education and heritage in Europe, and would be hard to explain to Europe’s voters, given the public support for such measures received from all sides of the debate so far.

This blog offers more detail on the situation so far, and sets out the case for avoiding this worst-case scenario.


The Good, The Bad and the Ugly. ACJ1, CC-BY-NC-SA https://www.flickr.com/photos/ajc1/4684652569The Good – Achievements So Far

The draft Directive already contains a lot of good. Starting from a reasonably positive base in September 2016, discussions among MEPs and Member States have led to improvements in provisions around text and data mining, teaching, preservation, and out-of-commerce-works – Articles 3-9.

If these elements of the Directive pass, EU citizens will:

  • Be able to engage much more easily in text and data mining. This will provide a significant boost to research into Artificial Intelligence in particular, at a time when Europe risks being left behind other countries who have been far more ready to update their legislation.
  • Have more opportunities to learn using digital tools, including in libraries. This will further democratise education, and help ensure that everyone can continue to learn throughout life.
  • Continue to enjoy access to Europe’s cultural heritage into the future, thanks to changes that will give libraries and cultural heritage institutions the clear right to take digital copies of books and other materials for preservation purposes.
  • Gain new access to works which are in-copyright but out-of-commerce, and so otherwise can only be found within the walls of libraries.

This is a good result, in and of itself. It will offer important clarity to libraries and cultural heritage institutions and allow them to fulfil their missions in the digital age. It will break down one of the most significant barriers to realising the potential of text and data mining, a Commission priority since 2012.

Moreover, given the EU’s own international obligations under the Berne Convention, it will not cause any unreasonable prejudice to authors. Instead, today’s authors will benefit from wider discovery of their work, including the rediscovery of works which are no longer in print. The authors of tomorrow will find it easier to read, study and innovate.

This is not to mention other elements of the text on the table that will provide additional rights to authors, including the possibility to reclaim rights and to benefit from greater transparency about revenues made on the basis of their work.

These provisions have enjoyed a large degree of consensus, with agreement relatively early on in discussions between Parliament and Council. Stakeholders from all sides of the discussion have been ready to signal their support for these steps, or at least their readiness to accept them.


The Bad – Sticking Points

However, it has long been clear that not all of the Directive is consensual. The two most contentious elements – Articles 11 and 13 – look to create new rights or rules for situations which are arguably specific to individual markets, and indeed individual providers – the situation of newspapers faced with GoogleNews, and of record companies faced with YouTube.

As has been argued repeatedly, the proposals on the table – a new right over very short fragments of text from newspapers, and an obligation on all online platforms to filter content uploaded by users – are likely to make the problem worse.

Not only will they strengthen the hand of the existing dominant players (who are best placed to negotiate with content producers, introduce filters or make payments), but they risk causing major collateral damage, for example to educational and scientific repositories run by libraries.

It is therefore unsurprising that there is so much disagreement about these articles.

Most recently, and just days after the agreement of a new Treaty between the countries, France and Germany disagreed about whether smaller internet platforms should be excused from the obligation to filter all user content for potential copyright infringement.

Even though this particular dispute has been agreed, there are many more still open, underlining how flawed the approach to these articles currently is.

In short, while there is support for effective ways of sustaining high quality journalism and curtailing illicit uses, the proposals on the table are not the answer.


The Ugly – The Nuclear Option

There are crucial meetings due in the coming days which aim to find a way forwards. Steps have been made to create some minor flexibilities in Articles 11 and 13, for example to reduce the burden on small platforms, as well as limited protections for the educational and scientific repositories that support open access and open educational resources.

Friends scene. Source: https://devrant.com/rants/1546587/this-will-happen-in-java-when-you-declare-the-class-with-wrong-nameHowever, there are already complaints from some who had previously supported Articles 11 and 13, who are unwilling to accept anything less than the highly flawed original proposals.

Most worryingly, these calls are accompanied by demands to reject the entire Directive.

This would be the worst of all worlds. All of the progress already made to date on Articles 3-9 would be at risk, despite already having been subject to consensus. The years of work that have gone into these would potentially be lost, and with it an opportunity to support clear public interest goals in Europe.

As an election approaches, it would be difficult to explain to voters why a flagship piece of legislation has been sunk, merely because there was disagreement on one part.

It is therefore time to reflect on the value of delaying those parts of the Directive which are clearly not yet mature, and proceeding with those that are. This would allow the European Union to chalk up a useful ‘win’.

Instead of rushed discussions now, a full and holistic discussion on how to achieve these goals, reviewing all relevant policy tools, is needed, and could be a useful job for the next Parliament.

Of Nuts and Sledgehammers: Why MEPs Should Choose their Tools Wisely in Copyright Reform

Graphic for sledgehammers and nutsThe European Parliament’s vote on the draft copyright directive next Wednesday is likely to be the last chance for transparent discussion on the substance of a reform that has been years in the making. It is also a last chance for libraries to reach out to and influence Members of the European Parliament.

A key message will be that European law-makers must choose wisely, and ensure that they are creating rules that are targeted, proportionate, and respect the public interest.

The Draft Copyright Directive

The last wide-ranging piece of EU copyright legislation dates to 2001. Since then, we have seen new technologies and expectations from users, dramatic evolutions in the market for music and media, and an explosion in the amount of copyrighted material produced every day online.

The draft Directive seeks to take stock of these changes, addressing questions around text and data mining, digital education, preservation, use of works which are no longer on sale, rights of press publishers and the obligations of content-sharing platforms, amongst other issues.

The debate has been intense, with a particular focus on Google and YouTube. It has, often, come across as a dramatic struggle between big technology companies and creators.

The problem with this approach is that tends to lead to dramatic solutions – sledgehammers to crack nuts. This blog illustrates just two areas where such dramatic solutions are being proposed, and the harm that they risk doing to libraries and their users.


Repositories are not YouTube

Perhaps the most contentious part of the Directive has been Article 13, which deals with the responsibility of content-sharing platforms to remove copyright-infringing materials uploaded by users.

While this covers commercial operations such as YouTube, other sites, such as educational and scientific repositories run by libraries and others also help people share their work. As such, they risk falling under the same rules.

For example, scientific repositories are a vital part of the infrastructure for open access. They host copies of research articles – often pre-print (i.e. not final) versions – allowing people who aren’t registered at the wealthiest universities or research institutions to have access. For doctors, individual researchers, and people in developing countries, this can be essential.

Educational repositories play a major role in spreading Open Educational Resources (OERs). These offer exciting possibilities for teachers to find and use materials which may be better tailored to their needs than traditional textbooks.

The repositories that host these materials are clearly working in the public interest, and are often hosted by libraries, education or research institutions. As concerns their size, resources, and objectives, they have little in common with YouTube.

However, the draft Directive risks treating them in the same way, placing the same regulations and responsibilities upon them. While YouTube can deal with this, it is hard to imagine repositories working on small budgets, and a strong aversion to legal risk, doing the same. See our blog on the risks around Article 13 and filtering for more.


Libraries are not Pirates

The desire to fight piracy of copyrighted content extends beyond Article 13. Elsewhere in the directive, organisations representing certain rightholders have made major efforts to impose restrictions on what libraries can do, claiming that this will help limit infringement.

For example, proposals on text and data mining (TDM) could make it very easy to restrict access to materials on the grounds of security, or force researchers to delete the datasets they create as part of the process. Such steps would create a major disincentive to invest time and effort in TDM.

Why do so when access to materials is uncertain, when the work that goes into structuring data will be lost, and when others will not be able to verify the results? Libraries already take care to respect copyright, and do not need further restrictions.

Similarly, there have been major efforts to prevent libraries from taking preservation copies of works held on third-party servers. In a digital world, this is the case for a growing share of what libraries offer their users. Excluding these eBooks, articles and other materials undermines a core mission of libraries, and increases the risk that these works in question being lost in future.

Finally, an amendment proposed to Article 6 of the directive would stop libraries using more than one exception at once. In practical terms, libraries would have to choose between taking a preservation copy of a work, carrying out text and data mining on it, or using it for teaching.

This would be a bizarre situation, with libraries forced to select which of their public interest missions they want to fulfil with works in their collections. It is also unnecessary, as whatever libraries do is still governed by copyright law, and in particular the obligation not to cause unjustified prejudice to rightholders. Libraries should not be forced to choose.


There are other areas where misguided rule-making risks doing more harm than good. Indeed, there is a strong argument that it is competition law, not copyright, that provides the best response to the market dominance of just a few major platforms.

While we will have to wait to see if Europe’s competition authorities act in this area. In the meanwhile – and particularly next Wednesday, it will be important to ensure that European law-makers choose their tools wisely.