How Filters fail (to meet the requirements of the DSM directive)

Sakkamettant l'eau au sir (filtre)
Three common failure modes of upload filters
Licentie

Article 17 of the DSM directive establishes that Online Content Sharing Service Providers (OCSSPs) are liable for copyright infringing uploads by their users unless they either obtain a license for the use of such content, or take a number of measures designed to prevent the availability of such content on their platforms. While the directive never explicitly talks about filters or automated content recognition (ACR) systems, it is assumed by all sides of the debate that, in order to meet this obligation, platforms have little choice but to implement ACR-based filtering systems that will scan all user uploads and block or remove uploads that contain works that have been flagged by their rightholders.

This de-facto requirement to implement upload filters is – by far – the most controversial aspect of the entire copyright directive and it continues to dominate the discussions about the implementation of Article 17 into national legislation.

In this context, it is important to remember that the use of such filters is not new and that their functioning can already be observed in practice. What is new, however, is the de-facto requirement for OCSSPs to implement filters as well as a number of requirements that OCSSPs need to meet to ensure that any measures (including filters) implemented by them are not infringing on the rights of users. This includes the requirement that any such measures “shall not result in the prevention of the availability of works or other subject matter uploaded by users, which do not infringe copyright and related rights, including where such works or other subject matter are covered by an exception or limitation“.

In other words, one of the most important contributions of the DSM directive is that, for the first time, it establishes conditions that need to be met by automated upload filters.

As we have argued many times before, these conditions present a very high hurdle for any technological solution to clear. The fact that upload filters are incapable of determining if a particular use of a copyrighted work is infringing or not has been established beyond any doubt. But that does not mean that the failure to assess the context is the only way that filters based on automated content recognition fail to meet the requirements established by the directive. In total there are at least three distinct ways how filters fail.

In the remainder of this post we will discuss these three failure modes based on examples collected by Techdirt in the course of a single week: removals caused by incorrect rights information, removals caused by the inability to recognise legitimate uses, and removals caused by the inability to accurately identify works.

Incorrect rights information

Incorrect rights information is probably the most common and best documented cause for the unjustified removal (or demonetisation) of works on YouTube.

ACR systems execute actions specified by whoever is recognised as the owner of a work. For the purposes of the ACR systems, the owner of a work is whoever claims to be the owner of the work and, unless there are conflicting ownership claims, there is no way to check the accuracy of such claims as there are no authoritative databases of ownership rights. As a result it is possible to claim ownership in public domain works (which no-one owns), in works that have been freely or widely licensed by their owners, or for any copyrighted work that has not already been claimed by someone else. Continue reading

A better way to implement Article 17? New German proposal to avoid overblocking

Jakobs ladder
A step-by-step concept against overblocking
Licentie

Back in April 2019, at the occasion of the final vote on the DSM Directive in the Council, the German Federal Government issued a statement, announcing that it intended to implement Article 17 with a focus on “preventing ‘upload filters’ wherever possible, ensuring freedom of expression and safeguarding user rights”. While the German Government has yet to produce an implementation proposal for Article 17, we may now have gotten a first insight in what such an implementation might look like. Late last month, the Kölner Forschungsstelle für Medienrecht at the University of Applied Sciences in Cologne, published a step-by-step concept for implementing Article 17 of the DSM Directive (“Stufenkonzept gegen Overblocking durch Uploadfilter“). 

The paper authored by Prof. Dr. Rolf Schwartmann and Prof. Dr. Christian-Henner Hentsch consists of an implementation proposal in the form of concrete legislative language. The objective of the authors seems to be to stay as close as possible to the objectives formulated in the German statement to the Council. What makes this proposal remarkable is that it is the first proposal (although not an official one) for implementing the Article 17 of the new Copyright Directive that does not consist of more or less literal transposition of Article 17 into national law (as it is the case in the French, Dutch and Belgian legislative proposals). In order to achieve the stated objective of preventing over-blocking by upload filters, the concept proposes a combination of Article 17 with Article 12 of the DSM Directive (which provides Member States the option to introduce Extended Collective Licenses). 

The implementation proposal contains modifications of three different acts: the Copyright Act (Urheberrechtsgesetz – UrhG), the Tele Media Act (Telemediengesetz – TMG) and the Collective Management Organisations Act (Verwertungsgesellschaftengesetz – VGG). Specifically the authors propose the following modifications: 

In the Copyright Act, they propose to add a new section to the article (§ 19a UrhG) that defines the act of communication to the public. The purpose of this addition is to include acts of giving the public access to copyright-protected user uploaded works by Online Content Service Providers (OCSSPs) in the definition of the existing making available to the public right. This establishes that, in principle, OCSSPs need authorisation from rightholders for such acts. The added section also includes the definition of OCSSPs, which is a literal transposition of the definition used in the DSM directive. 

The second addition to the Copyright Act is a new exception covering uses for the purpose of caricature, parody or pastiche by users of OCSSPs (§ 51a UrhG). Notably, this exception only applies in the context of sharing works via OCSSPs (which is highly unusual as copyright exceptions are usually not limited to specific services) and is conditional on remuneration to rightholders via collective management organisations. Continue reading

Research Librarians: New TDM exception can be undermined by technical blocking from publishers

De zanderij
Protect TDM from adverse effects of TPMs
Licentie

Today the Association of European Research Libraries (LIBER) issued a warning that the new Text and Data Mining (TDM) exceptions contained in the DSM directive can easily be undermined by technical blocking from publishers. LIBER has come to this conclusion based on the results of a survey on content blocking, carried out by LIBER’s Copyright & Legal Matters Working Group and the UK Libraries and Archives Copyright Alliance (LACA). The submissions to the survey so far confirm fears that Technical Protection Measures (TPM) can be abused by rightholders to limit the usefulness of the exceptions contained in the directive. 

Article 3 of the directive allows reproductions in the context of text and data mining for the purposes of scientific research made by research organisations and cultural heritage institutions. While the article allows rightholders to take measures to ensure the security and integrity of the networks, it does not allow them to prevent researchers from exercising their rights under the exception. It also requires rightholders to remove any TPMs that prevent researchers from exercising their rights but does not specify how quickly this has to happen. 

The survey conducted by LIBER shows that in practice it is difficult for researchers to get rightholders to remove TPMs or other access and use restrictions to content that research institutions have licensed. And in cases where such limitations are eventually removed the process is often long and resource intensive. All of this does not only frustrate efforts to text and data mining by individual researchers, but can also have negative consequences for entire institutions. 

The submissions shared with us by libraries and researchers from a mix of countries (some with TDM exceptions already and some without) show that:

  • Researchers are blocked from accessing many types of content. Journal articles were the most common type of content mentioned (44%) but eBooks, websites, databases and newspapers were also cited.
  • Content blocking takes, on average, nearly a month to resolve. Respondents reported that it took between 24 hours and 2-1/2 months to resolve the content blocking issue, with the mean time being 24 days. A fifth of survey respondents said they were only partially able to resolve the issue and 11% said it was never resolved.
  • Sanctions impact whole communities, not just individual researchers. Actions taken by publishers included 1) suspension of campus-wide access to paid for electronic subscriptions 2) threats to cut off access to content unless TDM was stopped 3) technically limiting downloads to one document only 4) a request for additional payments and 5) the introduction of captcha technology to frustrate data mining

The findings reported by LIBER are extremely relevant in the context of the ongoing implementations of the DSM directive in the EU member states. As we have noted in our implementation guide for articles 3 & 4 of the directive Member states should include in their national laws provisions requiring rightholders to remove TPMs and resolve lock-outs within a maximum period of 72 hours once reported.

In order to better make the case why such a requirement must be included in national implementations LIBER is continuing to collect evidence of abusive practices. If you work for an academic research institutions that has ran into similar issues in the past you can still contribute your experience to LIBER: 

If you or your organisation have ever been blocked from accessing a publisher’s servers for reasons you believe are related to data mining, fill out the survey. The survey can be answered anonymously and will remain open indefinitely. You can also send information on this issue to liberlegalworkinggroup@gmail.com”.

Article 17 stakeholder dialogue (day 6): Hitting a brick wall

A woman shouting into a man's ear-trumpetLicentie

This week saw the sixth meeting of the EU stakeholder dialogue on the implementation of Article 17 of the new copyright directive. This meeting was the first one where the question of how to reconcile the protection of user rights’ with automated filters based on technology that cannot assess the legality of the use of copyrighted works was explicitly on the agenda. From the perspective of the users’ organisations participating in the stakeholder dialogue this is the main question that the stakeholder dialogue (and the guidelines that the Commission will have to issue afterwards) needs to address. 

Unfortunately, Monday’s meeting did not result in a productive discussion about how to protect users’ rights. Proposals made by COMMUNIA and Studio71 on how to strike a balance between the rights of users and the interests of creators and other rightholders were largely ignored by a coalition of major rightholders from both the music and the audio-visual sectors. Working in concert, the representatives of the Hollywood studios (MPA), film producers (FIAFP), commercial television channels (ACT), major record labels (IFPI) and music publishers (ICMP) disputed the fact that there is a tension between protecting users rights and automated blocking, restated their conviction that Article 17 is only about empowering them versus the platforms, and suggested that users should simply trust that rightholders will not block free speech or other legitimate uses. In doing so they have made it clear that they want their interests to prevail at all cost, that users should not be trusted and that for them user rights are something that should exist at their discretion. 

This outcome leaves the European Commission in the difficult position to make sense of the input gathered throughout the previous six meetings and to outline a way forwards. Fortunately it seemed that the Commission is not willing to succumb to the unconstructive behaviour exhibited by rightholders and will take serious its task of finding a balance between users rights and the interests of different types of rightholders. 

A proposals for protecting users’ rights

So how could such a balance look like and what is at stake? One of the key insights that emerged from the previous rounds of the stakeholder dialogue is that even the most advanced content recognition technology is incapable of understanding the context in which works are used. This means that technology alone cannot make the determination if a use is lawful or not. Article 17 requires platforms to take measures to prevent the availability of content that rightholders want to keep off the sharing platforms and, at the same time, to ensure that legitimate uses (such as quotations or parodies) are not affected. This means that no matter how good it is at recognising content, ACR alone cannot meet the requirements of the directive. Continue reading

Article 17 stakeholder dialogue (day 5): It all depends

Copyright stakeholder dialogueLicentie

Last week Thursday saw the 5th meeting of the Commission’s stakeholder dialogue on Article 17 of the copyright directive. On paper this meeting was the first meeting where the stakeholders had the opportunity to discuss the actual provisions contained in Article 17 of the directive in order for the Commission “to gather evidence, views and suggestions that the services of the Commission can take into account” when preparing its guidelines on the application of Article 17. 

Contractual freedom über alles

In reality (recording available here), the 5th meeting was the meeting where it became clear that the vast majority of represented stakeholders is not interested in constructively contributing to the process and is instead participating in order to actively undermine the stakeholder process. Interventions made by commercial rights holders from the Music and AV sectors, as well as by platforms, focused on demanding maximum contractual freedom and arguing that the Commission’s guidelines should not contain any binding requirements. When asked to provide input for defining core concepts of the directive (such as “best efforts to obtain authorisation” and “best efforts to prevent availability”), most stakeholders limited their contribution to countless variations of “it depends” or theorising about “dynamic concepts”.

While there were some notable exceptions (apart from users organisations, collective management organisations and journalists’ organisations provided substantive input), it can hardly be surprising that both rightholders and big platforms have no interest in substantive guidelines that would offer meaningful safeguards for user rights. Continue reading

Article 17 stakeholder dialogue: What have we learned so far

A woman shouting into a man's ear-trumpetLicentie

This analysis was previously published in two instalments on the Kluver Copyright Blog (part 1, part 2).

As 2020 unfolds, the European Commission’s stakeholder dialogue pursuant to Article 17 of the Directive on Copyright in the Digital Single Market (CDSM directive) enters its third (and likely final) phase. After four meetings that focussed on gathering “an overview of the current market situation as regards licensing practices, tools used for online content management […] and related issues and concerns”, the next two (or more) meetings will finally deal with issues raised by the provisions in Article 17 of the CDSM directive. According to the Commission’s discussion paper for the meetings of 16 January and 10 February 2020, the objective of the third phase “is to gather evidence, views and suggestions that the services of the Commission can take into account in preparing the guidance pursuant to Article 17(10)”. 

In other words, after four meetings that have set the scene, the stakeholder dialogue will now address some of the thorny issues raised by Article 17. These include the key concepts like the best effort obligations to obtain authorisation and to prevent the availability of content (Article 17(4)), as well as the safeguards for legitimate uses of content (Article 17(7)) and the complaint and redress mechanisms available to users (Article 17(9)). In preparation for these forthcoming discussions, it is worth recapitulating what we have learned since the stakeholder dialogue kicked off in October of last year. 

Three takeaways from the stakeholder dialogue so far

After more than 25 hours of discussion (recordings of the four meetings can be found here: 1, 2, 3 and 4), there are three main insights that will likely have a substantial impact on the overall outcome of the stakeholder dialogue. These are the different motivations of different types of rightholders; the technical limitations of Automated Content Recognition (ACR) technologies; and the general lack of transparency with regards to current rights management practices. The first two of these are discussed in this post and the third will be covered in part 2 which will be published shortly. Continue reading

Article 17 stakeholder dialogue (day 4): it’s all about transparency

A woman shouting into a man's ear-trumpetLicentie

The fourth meeting of the Article 17 stakeholder dialogue took place in the last week before the holiday break. Just like the third meeting, this meeting was dedicated to (more or less) technical presentations on content management technologies and existing licensing practices. In total there were 12 presentations from platforms (Facebook, Seznam, Wattpad), providers of content management tools (Audible Magic, Ardito, Fifthfreedom, Smart protection), rightholders (GESAC, Universal Music Publishing, Bundesliga) as well as by BEUC and the Lumen database.

Filters are context-blind

The day’s presentations largely repeated the lines of arguments different stakeholders had presented during the previous meetings (a recording of the full meeting can be found here). Most notably all providers of content recognition technology confirmed that their technology does not go beyond simple matching of files and cannot understand the context in which a use takes place. Audible Magic summarised this in their opening statement:

Copyright exceptions require a high degree of intellectual judgement and an understanding and appreciation of context. We do not represent that any technology can solve this problem in an automated fashion. Ultimately these types of determinations must be handled by human judgement […]

As we have argued after the third meeting of the stakeholder dialogue, this is an unsurprising but significant insight as it means that current technology cannot be used to automatically block or remove content uploaded by users. 

Platforms don’t trust rightholders

The presentation given by Facebook about Facebook Rights Manager, its in-house content recognition tool, highlighted another problem that such tools are facing: One of the “main challenges” that Facebook is facing with its Facebook Rights Manager tool is that rightholders abuse the tool by claiming rights in works that they do not own. As a result Facebook only makes the most sensitive functionalities (such as automated blocking of uploaded content) available to a small group of carefully vetted trusted rightholders. Continue reading

Article 17 stakeholder dialogue (day 3): Filters do not meet the requirements of the directive

A woman shouting into a man's ear-trumpetLicentie

Last week’s third meeting of the Article 17 Stakeholder dialogue was the first one of what the Commission had referred to as the 2nd phase of the dialogue. After two meetings of introductory statements by various stakeholders (see our reports here and here), the third meeting consisted of a number of more in depth technical presentations on content recognition technologies and on existing licensing models (Video recording available here). 

The morning session saw presentations from three technology providers. YouTube presented its own Content ID system, PEX presented its platform independent attribution engine and finally Videntifier showed off its video and image matching technology

The biggest part of the discussion in the morning was centered around understanding the way YouTube’s content ID system works and how it relates to copyright (hint: it’s complicated). The overall impression that arose from the discussion is that very few participants actually understand how content ID works (and those who do, like the big record labels, don’t seem to be interested in talking about it). The fact that the Commission was among those asking questions to get a better understanding of the inner working of content ID is rather striking in the context that evidence based lawmaking was supposed to be one of the priorities of the Junker commission. So far the stakeholder dialogue seems more like an exercise in legislation based fact finding

While many aspects of Content ID remained opaque, one thing became clear though-out the three presentations: none of the presented technologies can do more than matching content in user uploads. None of the technologies presented can understand the context in which a use takes place and as a result they are incapable of detecting if a use is covered by an exception or not. In the words of the technology providers (lightly edited for clarity): Continue reading

Academic statement on safeguarding user freedoms in implementing Article 17

Optocht van professoren
51 academics say: if in doubt - don't filter
Licentie

Yesterday 51 leading European copyright scholars published a statement on “Safeguarding User Freedoms in Implementing Article 17 of the Copyright in the Digital Single Market Directive“. The statement is intended as input into the ongoing stakeholder dialogue. It focuses on the interplay between user rights and the filtering obligations established by Article 17. From the Kluwer Copyright blog:

Against this backdrop, a group of European academics (including the author of this post) has drafted a document with recommendations on user freedoms and safeguards included in Article 17 of the DSM Directive – namely in its paragraphs (7) and (9), to be read in the context of the aforementioned stakeholder dialogues. The recommendations are divided into three sections: on promoting licensing and limiting preventive measures; on exceptions and limitations in Article 17 (user freedoms); and on minimizing the risks of broad filtering and over-blocking (user safeguards). Despite the controversial nature of the topic, the recommendation has already been endorsed by around 50 European academics in this area, which is a testament to its balanced approach. Our intention is that these recommendations and interpretative guidelines are taken into consideration by the Commission, stakeholders, and Member States in their discussions on national implementations of Article 17 DSM Directive.

Based on a close reading of paragraphs 17(7) and 17(9), the academics show that Article 17 requires online platforms (OCSSPs in the language of the directive) to prioritise protecting users rights over blocking content. This statement provides strong support for our positions in the stakeholder dialogueContinue reading

Article 17 stakeholder dialogue (day 2): Filters, not licenses!

A woman shouting into a man's ear-trumpetLicentie

On Tuesday this week the participants of the stakeholder dialogue on Article 17 of the EU copyright directive convened in Brussels for the second meeting. After a first meeting that focussed on practices in the music, games and software sectors (see our report here), this week’s meeting focussed on the current situation in the audiovisual (AV) and publishing sectors. 

Hollywood: Article 17 is about filtering after all

The meeting kicked off with a long series of statements from the many different rightholders in the AV sector (see the video recording here). The assembled sector representatives made it clear that from their perspective Article 17 is welcome (as it clarifies that online platforms need to obtain licenses for the works uploaded by their users) but that they are not interested in widely licensing AV works to UGC platforms and would instead focus on the blocking and removing of unlicensed content via the upload filtering mechanisms introduced by Article 17. 

This approach is the logical consequence of the predominant business model in the AV sector which relies on exclusive licensing to selected outlets (Cinema, TV, VOD platforms). It directly contradicts the music industries’ narrative from the first meeting that Article 17 is about licensing and not about blocking access – as in the case of music general availability is crucial. Representatives of the AV industry made it very clear that they would fight any attempts at non-voluntary licensing and that they would also fight against effective protection for user rights under exceptions and licensing (see for example the statement issued by the Motion Picture Association starting at 10:41:44 of the video recording). These initial statements make it clear that the AV industry does indeed look at Article 17 as an instrument to limit freedom of expression and reuse and will likely use the stakeholder dialogue to bend the article further in this direction. Continue reading