Blogs | Information democracy | Open internet and inclusive technology | Transparency

Copyright stakeholder dialogues: Filters can’t understand context

By EDRi · January 15, 2020

On 16 December 2019, the European Commission held the fourth meeting of the Copyright Directive Article 17 stakeholder dialogues. During the “first phase”, meetings focused on the practices in different industries such as music, games, software, audiovisual and publishing. This meeting was the last of what the Commission called the “second phase”, where meetings were focused on technical presentations on content management technologies and existing licensing practices.

During this fourth meeting, presentations were given by platforms (Facebook, Seznam, Wattpad), providers of content management tools (Audible Magic, Ardito, Fifthfreedom, Smart protection), rightsholders (European Grouping of Societies of Authors and Composers – GESAC, Universal Music Publishing, Bundesliga) as well as by consumer group BEUC and the Lumen database.

Say it again louder for the people in the back: Filters cannot understand context

After Youtube’s Content ID presentation during the third meeting, Facebook’s Rights Management tool presentation reiterated what civil society has been repeating during the entire duration of the copyright debates: filtering tools cannot understand context. Content recognition technologies are only capable of matching files and cannot recognise copyright exceptions such as caricature or parody.

This argument has now been clearly and repeatedly laid out to the European Commission by both civil society organisations and providers of content recognition technology. We would therefore expect that the Commission’s guidelines will take this into account and recommend that filters should not be used to automatically block or remove uploaded content.

A lack of trust

As the meetings usually help revive old divisions between stakeholders, it also tells us about new ones. Facebook’s Rights Management tool pointed out that one of their biggest issues was the misuse of the tool by the rightsholders who claim rights on work they do not own. As a result, not every rightsholder get access to the same tools. Some tools such as automated actions are limited or reserved for what the provider calls “trusted rightsholders”.

On the other side, rightsholders such as GESAC have criticised the way they are treated by the big platforms such as YouTube. In particular, they highlighted the categorisation made by the content recognition tools which can lead to loss of revenue. Indeed, rightsholders sometimes have no choice but to use tools created and controlled by big platforms with their own opaque rules, and therefore emphasised the need for transparency and accuracy of the information on the way platforms like YouTube operate with content whose rights they own.

Transparency is key

With the aim of understanding the management practices of copyright-protected content, quantitative information is crucial. Faced with the issue of filters, content recognition providers said they have been relying on redress mechanisms and human judgment. But when asked for factual information on the functioning of their practices, no number or percentage was available. It is therefore impossible to understand the necessity, proportionality or efficiency of the use of automated content recognition tools.

According the Article 17(10) of the Copyright Directive, which provides the basis for the ongoing stakeholder dialogue, “users’ organisations shall have access to adequate information from online content-sharing service providers on the functioning of their practices with regard to paragraph 4.”

After four meetings and still lacking such information from companies, civil society organisations participating in the dialogue decided to send a request for information to the European Commission. We hope that the Commission will be able to gather such factual information from platforms so that the ongoing dialogue can lead to an evidence-based outcome.

As part of these transparency needs, EDRi also signed an open letter asking the Commission to share the draft guidelines they will produce at the end of the dialogue. In the letter, we asked that the guidelines should also be opened to a consultation with the participants of the stakeholder dialogues and to the broader public, to seek feedback on whether the document can be further improved to ensure compliance with the Charter of Fundamental Rights of the EU.

What’s next?

The next stakeholder dialogue meeting will be held on 16 January and will open the “third phase” of consultation, which will focus on the practicality of Article 17. The Commission already sent out the agenda, and the topics covered on 16 January will be authorisations, notices and the notion of “best efforts”, while the following session on 10 February will cover safeguards and redress mechanisms.

EU copyright dialogues: The next battleground to prevent upload filters (18.10.2019)
https://edri.org/eu-copyright-dialogues-the-next-battleground-to-prevent-upload-filters/

NGOs call to ensure fundamental rights in copyright implementation (20.05.2019)
https://edri.org/ngos-call-to-ensure-fundamental-rights-in-copyright-implementation/

Copyright: Open letter asking for transparency in implementing guidelines (15.01.2020)
https://edri.org/copyright-open-letter-asking-for-transparency-in-implementing-guidelines

(Contribution by Laureline Lemoine, EDRi)