Suchergebnisse

AbstractMachine learning, a subfield of artificial intelligence (AI), relies on large corpora of data as input for learning algorithms, resulting in trained models that can perform a variety of tasks. While data or information are not subject matter within copyright law, almost all materials used to construct corpora for machine learning are protected by copyright law: texts, images, videos, and so on. There are global policy moves to address the copyright implications of machine learning, in particular in the context of so-called "foundation models" that underpin generative AI. This paper takes a step back, exploring empirically three technological settings through detailed case studies. We set out the established industry methodology of a lifecycle of AI (collecting data, organising data, model training, model operation) to arrive at descriptions suitable for legal analysis. This will allow an assessment of the challenges for a harmonisation of rights, exceptions and disclosure under EU copyright law. The three case studies are:



Machine learning for scientific purposes, in the context of a study of regional short-term letting markets;





Natural Language Processing (NLP), in the context of large language models;





Computer vision, in the context of content moderation of images.


We find that the nature and quality of data corpora at the input stage is central to the lifecycle of machine learning. Because of the uncertain legal status of data collection and processing, combined with the competitive advantage gained by firms not disclosing technological advances, the inputs of the models deployed are often unknown. Moreover, the "lawful access" requirement of the EU exception for text and data mining may turn the exception into a decision by rightholders to allow machine learning in the context of their decision to allow access. We assess policy interventions at EU level, seeking to clarify the legal status of input data via copyright exceptions, opt-outs or the forced disclosure of copyright materials. We find that the likely result is a fully copyright-licensed environment of machine learning that may have problematic effects for the structure of industry, innovation and scientific research.

Platforms have emerged as a new kind of regulatory object over a short period of time. There is accelerating global regulatory competition to conceptualise and govern online platforms in response to social, economic and political discontent – articulated in terms such as 'fake news', 'online harms' or 'dark patterns'. In this paper, we empirically map the emergence of the regulatory field of platform regulation in the UK. We focus on the 18-month period between September 2018 and February 2020 which saw an upsurge of regulatory activism. Through a legal-empirical content analysis of eight official reports issued by the UK government, parliamentary committees and regulatory agencies, we (1) code over 80 distinct online harms to which regulation is being asked to respond; we (2) identify eight areas of law referred in the reports (data protection and privacy, competition, education, media and broadcasting, consumer protection, tax law and financial regulation, intellectual property law, security law); we (3) analyse nine agencies mentioned in the reports for their statutory and accountability status in law, and identify their centrality in the regulatory network; we (4) assess their regulatory powers (advisory, investigatory, enforcement); and the regulatory toolbox of potential measures ascribed to agencies; we (5) quantify the number of mentions platform companies received in the reports analysed. We find that Ofcom (the communications regulator) and the CMA (the Competition and Markets Authority) are the most central actors in the regulatory field, with the Information Commissioner (the data regulator) following close behind. We find that security- and terrorism-related interventions remain particularly obscure and hard to capture with a socio-legal analysis of public documents. We find that the political focus is overwhelmingly on a handful of US multinational companies. Just two companies, Google and Facebook, account for three-quarters of the references made to firms in the documents we examined. Six Chinese ...

The Directive on Copyright in the Digital Single Market has been a subject of heated and highly polarised debate, and an object of intense lobbying from the outset. It grasped the attention of a multitude of stakeholders, including tech companies, publishers, platforms, creators and SMEs, and urged thousands of people to go out on the streets in a sign of protest against what they believed was the "end of the internet as we know it". The debate was often emotional, and involved such terms as "meme ban", "censorship", "upload filters", "link tax", or a puzzling "value gap". However, amongst those emotive catchphrases lies a foundational discussion on the purposes of copyright law, and how its relationship with artists, technology, media, news, culture and citizenship unfolded. In this presentation, we investigate how discourse developed during the negotiation phase of the Directive, focusing on the most controversial provisions: draft Articles 11 (press publishers' right) and 13 (platform liability). Focusing on the period between publication of the proposal by the Commission in September 2016, and the adoption of the Directive by Parliament and Council in March/April 2019, we juxtapose these changes with an analysis of (1) parliamentary debates, (2) press releases by the Commission, Parliament and Council, and (3) 80 stakeholder submissions that sought to shape the evolving legislation. Through discourse analysis, we uncover four topoi that appear to dominate the debate: (a) Technocratic (responding to tech development by updating the copyright framework), (b) Value gap (the redistribution of revenues to benefit creators and producers), (c) Internet freedoms (freedom of expression and user interests), and (d) European (the promotion and protection of European culture and identity). Finally, we show that changes in the Directive's text can be associated with the appearance and evolution of the discourse. We show that changes in the draft legislation can be associated with the appearance and evolution of the four topoi in the debates, but that changes in the proposed legal language tend to be obfuscatory, rather than addressing the issues. Controversial language such as "content identification" (associated with filtering) was removed and safeguards were offered, but many of these changes remained meaningless (in law) or open to different implementations.

© 2020 The Authors. The Modern Law Review published by John Wiley & Sons Ltd on behalf of Modern Law Review Limited. The juridification of the European policy process is increasingly fragile, and little understood. This study develops a novel methodology to investigate the influence of Member States on the rulings of the Court of Justice of the European Union (CJEU). The focus is on the domain of copyright law which has seen a dramatic escalation of preliminary references to the Court, indicating a normative void. Examining 170 documents relating to 42 cases registered between 1998 and 2015, we measure empirically the impact of submissions by Member States and the European Commission on the interpretation of copyright concepts. We show that France is the most influential country by some distance, both in terms of the number of interventions (an 'investment' in policy) and in terms of persuasive power (arguments adopted by the Court). The evidence also suggests that the departure of the UK from EU litigation will disturb the delicate balance of CJEU jurisprudence.

The Directive on Copyright in the Digital Single Market has been a subject of heated and highly polarised debate, and an object of intense lobbying from the outset. It grasped the attention of a multitude of stakeholders, including tech companies, publishers, platforms, creators and SMEs, and urged thousands of people to go out on the streets in a sign of protest against what they believed was the "end of the internet as we know it". The debate was often emotional, and involved such terms as "meme ban", "censorship", "upload filters", "link tax", or a puzzling "value gap". However, amongst those emotive catchphrases lies a foundational discussion on the purposes of copyright law, and how its relationship with artists, technology, media, news, culture and citizenship unfolded. In this presentation, we investigate how discourse developed during the negotiation phase of the Directive, focusing on the most controversial provisions: draft Articles 11 (press publishers' right) and 13 (platform liability). Focusing on the period between publication of the proposal by the Commission in September 2016, and the adoption of the Directive by Parliament and Council in March/April 2019, we juxtapose these changes with an analysis of (1) parliamentary debates, (2) press releases by the Commission, Parliament and Council, and (3) 80 stakeholder submissions that sought to shape the evolving legislation. Through discourse analysis, we uncover four topoi that appear to dominate the debate: (a) Technocratic (responding to tech development by updating the copyright framework), (b) Value gap (the redistribution of revenues to benefit creators and producers), (c) Internet freedoms (freedom of expression and user interests), and (d) European (the promotion and protection of European culture and identity). Finally, we show that changes in the Directive's text can be associated with the appearance and evolution of the discourse. We show that changes in the draft legislation can be associated with the appearance and evolution of the four ...

The juridification of the European policy process is increasingly fragile, and little understood. This study develops a novel methodology to investigate the influence of Member States on the rulings of the Court of Justice of the European Union (CJEU). The focus is on the domain of copyright law which has seen a dramatic escalation of preliminary references to the Court, indicating a normative void. Examining 170 documents relating to 42 cases registered between 1998 and 2015, we measure empirically the impact of submissions by Member States and the European Commission on the interpretation of copyright concepts. We show that France is the most influential country by some distance, both in terms of the number of interventions (an 'investment' in policy) and in terms of persuasive power (arguments adopted by the Court). The evidence also suggests that the departure of the UK from EU litigation will disturb the delicate balance of CJEU jurisprudence.

The Court of Justice of the European Union (CJEU) has been suspected of carrying out a harmonising agenda over and beyond the conventional law-interpreting function of the judiciary. In relation to the development of a EU copyright law, the Court has seen a dramatic increase in activity, with 6 cases filed in the 10 years following the Phil Collins case of 1992, 6 cases filed in the 5 years between 2002 and 2006, and 26 cases in the 5 years between 2007 and 2011. This study aims to investigate empirically two theories in relation to the development of EU copyright law: (i) that the Court has failed to develop a coherent copyright jurisprudence (lacking domain expertise, copyright specific reasoning, and predictability); (ii) that the Court has pursued an activist, harmonising agenda (resorting to teleological interpretation of European law rather than – less discretionary – semantic and systematic legal approaches). The findings of the study confirm the former, and qualify the latter.

The Court of Justice of the European Union has seen a dramatic and controversial increase in copyright cases during the last decade. This study investigates empirically two claims: (i) that the Court has failed to develop a coherent copyright jurisprudence (lacking domain expertise, copyright specific reasoning, and predictability); (ii) that the Court has pursued an activist, harmonising agenda (resorting to teleological interpretation of European law).We analyse the allocation of copyright and database right cases by Chambers of the Court, Advocate General (AG) and Reporting Judge, and investigate the biographical background of the Judges and AGs sitting. We trace patterns of reasoning in the Court's approach through quantitative content analysis. We identify the legal topoi that are employed in the opinions and decisions, and then link the occurrence of these topoi to the outcome of each case. The results show that private law and in particular intellectual property law expertise is almost entirely missing from the Court. However, we find that the Court has developed a mechanism for enabling judicial learning through the systematic assignment of cases to certain Judges and AGs. We also find that the Court has developed a "fair balance" topos linked to judge Malenovský (rapporteur on 24 out of 40 copyright cases) that does not predict an agenda of upward harmonisation, with about half of judgements narrowing rather than widening the scope of copyright protection.

The Court of Justice of the European Union (ECJ) has been suspected of carrying out a harmonising agenda over and beyond the conventional law-interpreting function of the judiciary. This study aims to investigate empirically two theories in relation to the development of EU copyright law: (i) that the Court has failed to develop a coherent copyright jurisprudence (lacking domain expertise, copyright specific reasoning, and predictability); (ii) that the Court has pursued an activist, harmonising agenda (resorting to teleological interpretation of European law rather than – less discretionary – semantic and systematic legal approaches). We have collected two data sets relating to all ECJ copyright and database cases up to Svensson (February 2014): (1) Statistics about the allocation of cases to chambers, the composition of chambers, the Judge Rapporteur, and Advocate General (including coding of the professional background of the personnel); (2) Content analysis of argumentative patterns in the decisions themselves, using a qualitative coding technique. Studying the relationship between (1) and (2) allows us to identify links between certain Chambers/ Court members and legal approaches, over time, and by subject. These shed light on the internal workings of the court, and also enable us to explore theories about the nature of ECJ jurisprudence. The analysis shows that private law and in particular intellectual property law expertise is almost entirely missing from the Court. However, we find that the Court has developed a mechanism for enabling judicial learning through the systematic assignment of cases to certain Judges and AGs. We also find that the Court has developed a "fair balance" topos linked to Judge Malenovský (rapporteur on 24 out of 40 copyright cases) that does not predict an agenda of upward harmonisation, with about half of judgments narrowing rather than widening the scope of copyright protection.