Editorial

While producing this issue of Stockholm Intellectual Property Law Review, we are just about to say goodbye to 2022 and welcome 2023, a year we hope will be full of growth and repose from the destruction and demise of these last few years. After our previous issue celebrating the founder of the European IP Law masters programme, Professor Marianne Levin, we are excited to bring you this issue with several thought provoking articles that test the bounds of society, culture, art, and technology on IP.

In modern times, one of the most present and enduring technological advancements that have had an impact on how we think about intellectual property (IP) is artificial intelligence (AI).

AI is without a doubt both a trouble maker and simulator testing the ability for IP laws to adapt to new applications and technology. From Dabus with its question as to whether AI-created inventions can be granted patent protection to the next big thing on the legality of  the use of training data in developing AI, AI seems to be a never ending stream of thought provocation.

Microsoft, initially a strong opponent of the open source movement, gradually ‘joined’ in and even purchased GitHub in 2018, a platform that allows programmers to develop and store their open access code.

In 2022, a new kind of AI technology was released by Microsoft that can generate its own computer code. Github Copilot, as it is named, speeds up the work of professional programmers by suggesting to them ready-made blocks of computer code that they could instantly incorporate in their work. Programmers seemed to love this new tool! Well, at least most of them did. Matthew Butterick, a programmer that is also a designer, an author and a lawyer was not so happy with this latest Microsoft project and decided to file a class-action lawsuit against Microsoft and other high-profile companies that have participated in the GitHub Copilot project.^[1]

Butterick claims that GitHub Copilot was only possible due to the availability of billions of lines of computer code that was made available on the internet.^[2] It is the work of the computer programmers who spend years writing the code, and who now are not acknowledged in any way, that in fact made GitHub Copilot.^[3] This seems to be the first legal action concerning ‘AI training data’, the most important aspect  of constructing an AI system.

Of course Butterick’s lawsuit is not the first time concerns as to the status and practices of ‘training data’ are raised. Artists, authors, programmers and composers have during the past few years raised their concerns that companies and AI researchers actually used their work without their consent and without any much less adequate remuneration. The AI applications built from ‘training data’ are extremely broad ranging from art generators to speech recognition systems and even automated cars.

Microsoft has claimed that the use of existing code to train AI is done under the legal doctrine of ‘fair use’, and this argument is definitely not a new one, however it is one that has not yet been tested in the US courts (or elsewhere).

Prior to GitHub Copilot in 2020, OpenAI (an AI lab run by Microsoft), released an AI system called GPT-3. This is an AI system that has been trained using vast amounts of digital text, thousands of books, Wikipedia articles, chat logs and other data available online. The system learned to predict what word is to come in a sentence. Gradually, it began completing the thoughts of an author by suggesting whole paragraphs, then evolving to provide whole pages, poems, articles and speeches. In fact, it could even write computer programs.

OpenAI then took the project a step further training a new system, OpenAI Codex, that was specifically trained with computer programmes. OpenAI Codex then gradually led to GitHub Copilot. And while GitHub Copilot produces only simple code that requires the contribution of a programmer in order for it to be usable, we know developments run fast.

Butterick is not only concerned with issues of acknowledgement for the authors but also with the impact this AI application will have on the global community of programmers.^[4] Being part of the open source community he claims that open source software stands today for the most important tech applications we use in everyday life.^[5] While open source code is shared freely, this sharing has its legal basis on licenses designed to ensure that it is used in ways that would benefit the community of programmers. According to Butterick Microsoft has violated the terms of the licenses and in fact, if GitHub Copilot continues to improve it will make open source programmers obsolete.^[6]

Of interest here is that Butterick does not base the lawsuit on copyright infringement but instead concentrates on claims that he argues are not subject to a fair use defense.^[7] He argues that companies have violated GitHub’s terms of service and privacy policies while also violating federal law that requires companies to display copyright information when they make use of the material.^[8]

This lawsuit is representative of the challenges AI poses on the IP system as we know it. And makes it clear that the use of training data is without a doubt important from an IP law perspective and necessary to ponder.

This issue of SIPLR discusses several issues that are part directly or indirectly of the challenges brought on the copyright system by creative processes and technological and societal changes, and how these should be managed.

In his article, A Reflection on the Cultural Significance of the Protection of Classics, Martin Fredriksson discusses the fall and rise of the protection of classics in Swedish legislation. Having the Nordfront case as a starting point, he walks you through his alluring analysis on the meaning of § 51 of the Swedish Copyright Act and in particular the meaning of violation of the ‘interests of spiritual cultivation’.

Following this, Marina Katrakazi, Panagiota Koltsida, Eleni Toli, and Prodromos Tsiavos present in their article License Clearance Tool: A holistic open IP and open innovation practices among research communities  the practical applications of a License Clearance mechanism (LCT). As explained in detail in the article, an LCT focuses on automating the clearance of Intellectual Property Rights (IPR) by ensuring the compatibility among different licenses included in the same resource. The article delves into the growth of the open source world and the value added by the LCT system.

In the third article, Balancing Article 17 CDSMD and the Freedom of Expression, Finn Hümmer discusses recent case law, national and EU legislation and offers a timely contribution to the debate on the congruity of Article 17 of the DSM Directive with the fundamental right to freedom of expression and information.

Finally, in an engaging  contribution, The Machinery of Creation. Oulipo Poetry, Copyright & Rules of Constrain, Kathy Bowrey and Janet Bi Li Chan present the creative process of Oulipo poetry and analyse the impact of copyright law on this form of expression. The authors eloquently reconcile love for the art of poetry with copyright law’s influence on the art. Concluding with a unique sentiment, ‘Still, copyright law has quite a lot in common with Oulipo. Obvious similarities include that legal reasoning is often imagined as a semi-closed machine, where language choices produce new meaning. But there is a foundational plagiarism in copyright – the reproduction of a humanist authorial beneficiary of law used to anchor the legal machinery of infringement. This confinement means that copyright is unable to properly converse with artists or poets about a key difference between copyright and Oulipo. Law suppresses the cyborg in all creation.’

This issue of the SIPLR is produced by a new group of student editors and a new student editor in chief, all of them working toward their masters in European Intellectual Property Law at Stockholm University! Without their contribution the production of this issue would not be possible.

We hope you enjoy reading this thought provoking  issue 2022 (2) of the SIPLR!

Frantzeska Papadopoulou,

Professor of Intellectual Property Law, Law Faculty, Stockholm University

^[1] Complaint, Matthew Butterick v Github, Inc. et al. (N.D. Cal.) Case 3:22-cv-06823, Nov 3, 2022, copy available <https://githubcopilotlitigation.com/pdf/06823/1-0-github_complaint.pdf> accessed 23 January 2023. ^[2] ibid [84].  ^[3] ibid [192] – [195]. ^[4] ibid [164]. ^[5] ibid [106]. ^[6] See Cade Metz, ‘Lawsuit Takes Aim at the Way A.I. Is Built’ The New York Times (New York, 23 November 2022) <https://www.nytimes.com/2022/11/23/technology/copilot-microsoft-ai-lawsuit.html> accessed 23 January 2023. ^[7] Complaint, Butterick v Github (N.D. Cal.) Case 3:22-cv-06823 [85]. ^[8] ibid.

ABSTRACT

This article applies a cultural perspective on § 51 of the Swedish Copyright Act, which prohibits the rendering of works in the public domain ‘in a way that offends the interests of spiritual cultivation’ (SFS 1960:729). This so called ‘protection of classics’ was formulated in the 1950s to protect classical works against derogatory interpretations, such as popular cultural adaptions. § 51 has rarely been applied, but in 2021 it was for the first time tried in court as the nationalist website Nordfront was accused of violating §51 by publishing works by three prominent romanticist poets in a context bordering on hate speech. The court ruled that the publication was not a violation of § 51, which calls the future of the protection of classics into question. Even though §51 might soon be obsolete, it raises a number of questions regarding the relation between law and culture. This article discusses what the protection of classics and the Nordfront case can tell us about cultural change in postwar Sweden if it is approached as a cultural rather than a legal text and studied not primarily as a legislative process but as a process of meaning making. The article makes no attempts to conduct such an analysis but rather aims to introduce the perspective and present preliminary reflections on how the formulation and use of protection of classics reflects changing conceptions of cultural norms and values.

ABSTRACT

Open Science (OS) movement, remote collaboration among research communities and the increased quantity of new content, data and resources, have made it clear that traditional licensing schemes require new tools that would combine technical and legal features, as well as techno-licensing tools. In fact, the diversity of open licenses deprived from standardization frequently leads to situations where more than one open licenses with different or conflicting terms apply at the same time, and hence it gives rise to license compatibility concerns. This creates a legal uncertainty that may discourage authors, scientists, and researchers from releasing their work under an open license. In this paper, we identify legal and technological barriers that pose a challenge in adopting open science practices; thereafter, we present a new tool, named License Clearance Tool (LCT), which has been developed by the Athena RC (Greece) as part of the National Initiatives for Open Science in Europe – NI4OS-Europe (https://ni4os.eu/), a European project that contributes to the European Open Science Cloud (EOSC) by supporting its activities in Southeast Europe. LCT is an open-source tool, which provides a holistic approach addressing IP issues. LCT focuses on automating the clearance of Intellectual Property Rights (IPR) by ensuring the compatibility among different licenses included in the same resource and assists users on the selection of the most suitable license by providing a content summary of them with respect to permissions, prohibitions, and obligations in relation to the user needs. It is intended to support mainly researchers and non-legal experts in general to publish in FAIR/open modes.

ABSTRACT

Accompanied by massive protests driven by concerns about the introduction of upload filters, Art. 17 of Directive (EU) 2019/790 on copyright in the Digital Single Market (CDSMD) came into force in 2019. Since then, a lot has happened: Poland filed an action for annulment, whose ruling was announced by the CJEU on 26 April 2022, Member States delivered transpositions with diverging approaches, and a political agreement on the Digital Services Act, which contains numerous regulations that are potentially also applicable to online platforms as addressees of Art. 17 CDSMD, was reached on 23 April 2022. With all these regulations, the question remains open as to how it can be ensured that Art. 17 CDSMD is compatible with the fundamental right to freedom of expression and information, enshrined in Art. 11 of the Charter of Fundamental Rights. In order to answer this question, the origin of Art. 17 CDSMD and its ratio, its liability mechanism and its impact on the fundamental right will first be examined. In a further step, the safeguards laid down in Art. 17 CDSMD are analysed and the effects of the current ruling on national implementations are discussed. Finally, the focus is on the German approach, which contains farther-reaching ex ante safeguards and is considered as a model for further implementations, especially after the judgement of the CJEU.

ABSTRACT

This article explores the double burden of creative regulation – the aesthetic restrictions artists choose and their interaction with copyright rules, using the example of oulipo, a constraint-based creative practice. Part One explains the Oulipo movement. Oulipo technique is then demonstrated in new poems by artist and poet Janet Bi Li Chan, based on existing works, Franny Choi’s Turing Test and Tracy K Smith’s Sci-FI, applying word substitution (N + 7), erasure or blackout technique, and remixing. Part Two applies a copyright law reading to the new poems. We show how legislative frameworks measure all creators – regardless of artistic self identity and process – as if they were humanist authors. But in application tests also produce far more surprises than might be expected if law is conceived of as rule-based constraint. Part Three applies Oulipo techniques to key articles of the Berne Convention that permit artistic licence: Art 9 Right of Reproduction; Art 10 Certain Free Uses of Works; and Art 10bis Further Possible Free Uses of Works. These new Berne poems highlight the prescriptive face copyright law presents to creators. We argue it is the emotional resonance of copyright, rather than its technicality, that primarily impacts creative practice. We conclude that law reproduces an idealised imaginary of a humanist author to measure creative transgression and this confinement means that copyright is unable to properly converse with artists or poets. Law suppresses the cyborg in all creation.