In new research, Christian Peukert and Margaritha Windisch review how copyright laws and practices have evolved to adapt to new technologies and discuss the various issues scholars and policymakers must address as copyright law is once again forced to adapt to the emergence of artificial intelligence.
Intellectual property rights underpin how economies promote and organize innovation and steer the diffusion of knowledge. Copyright law, in particular, must constantly evolve to keep up with emerging technologies, how these technologies alter the various interests of creators, consumers, and intermediaries, and how these interests interact in different creative industries. Although earlier inventions, such as the printing press and the camera, enabled cheaper and faster creation and copying of works, copyright law was able to cope with the changes: for most of copyright history, the technology, storage, and distribution were still analog, which made the speed of production and reproduction manageable for copyright enforcement. The emerging transition to digital technologies, beginning with the introduction of the microchip in 1959 and personal computer in 1981, and then considerably accelerated by the internet, has challenged stakeholders in the copyright system on how to respond to the accelerated capabilities to create, distribute, and copy creative works.
Similar to patent law, copyright law tries to solve a public goods problem (why someone should provide a good that the public will be able to freely access) by granting an exclusive bundle of rights to creators to control the reproduction and distribution. For example, the author of a book can decide who can publish it, make copies, or create derivative works, such as a movie. By having these rights, the author can recoup investments made. In most countries, the period of exclusive rights lasts for the life of the author plus 70 years. The rationale is clear: without such protection, creative works would be underproduced because anyone could use it without the creator receiving compensation. Which is not to say money is the sole driving force behind creativity, as ample empirical and anecdotal evidence suggests. The personality rights of authors—so-called moral rights, such as the right to attribution—play a significant role, too.
Copyright in the age of the internet
Since creative activities often build (partly) upon existing works, licensing agreements play an essential role in the copyright system—think of music recordings that are often reused and recombined for sampling and remixing. From the late 1990s onward, emerging technologies enabled the digitization of works such as music, movies, and books and, with it, the ability to copy existing works without any additional cost and without lowering the quality. The digitization of works facilitated reuse, but existing copyrights and a lack of available licensing models often become an obstacle for creators and end-users. For example, a study shows that copyright restrictions drastically reduce citations to copyrighted works on Wikipedia, which results in lower-quality articles (as measured by page visits). Empirical evidence shows that the monopoly power created by copyrights results in higher prices and lower availability for copyrighted works compared to those in the public domain. Once the copyright expires, the lower costs of accessing the work encourage follow-on innovation.
The high costs to licensing copyrighted works encourages copyright infringement, and the rise of the internet made this easier. Consequently, the lower effort and costs of unlawful accessing, downloading, and distributing of copyrighted works have led to massive-scale copyright infringement on the demand side, affecting the revenues of various creative industries. The cases of Napster and other peer-to-peer file-sharing networks enabling unlicensed music distribution in the early 2000s illustrate the difficulty of identifying, prosecuting, and obtaining damages from massive-scale online copyright infringement. Some evidence suggests that the impact to creators of losing sales to pirated distribution of their works is negligible or even positive because of word-of-mouth effects. In contrast, other studies suggest a loss of music, film, and software sales up to two-thirds. In response to decreasing revenues, creators and firms have sought income in alternative business models, such as live concerts, freemium subscription services for software, music, and video, and by changing their IP strategies.
Social media and copyright
The more recent rise of social media has posed a particular quandary for copyright, with most digital content, including copyrighted music and film, made available today via intermediary online platforms, such as YouTube and TikTok. Safe harbor provisions in both the United States and the European Union provide a legal framework to exempt platforms from liability under certain conditions, and (automated) Notice and Takedown (N&TD) regimes implemented by the platforms provide copyright holders with an easier way to resolve online copyright infringement than filing lawsuits. However, the determination of copyright infringement has long been marked by confusion, especially in cases of reuse and the assessment of substantial similarity between two works. While exceptions to copyright—such as the fair use doctrine in the U.S. and exhaustive catalogs in European countries—aim to promote the progress of the arts and sciences, automated N&TD effectively allows online platforms to replace the legal use of copyrighted material with their own rules, which can lead to incorrect labeling of fair-use content and the removal of content that does not involve copyright at all. The problem is that automated systems still lack an understanding of context, and the non-transparent nature of automatic filter mechanisms creates uncertainty, which can hinder follow-on innovation by affecting creative decisions and distribution strategies. As a result, creators may innovate around existing ideas rather than seeking a license or decide not to innovate at all.
To overcome problems of copyright, online platforms implementing automated N&TD systems have, over the last decade, switched the focus from pure enforcement to fostering a digital licensing market. For example, around the time YouTube launched the affiliate program in 2009, in which the platform agreed to share advertising revenue with contributors, YouTube introduced its “Content ID” system, where rightsholders are notified of allegedly infringing content and can choose to block the video or monetize the content by sharing advertising revenue. YouTube reports that over 90% of rightsholders choose monetizing, which has accumulated to $7.5bn in advertising revenue paid to rightsholders from claims through the system since 2016, suggesting that it has become feasible to license millions of derivative works with negligible cost.
The introduction of on-demand streaming services has further reshaped revenue models for rightsholders. Platforms like Spotify, YouTube, and Netflix offer extensive content catalogs through subscription or ad-supported models. This raises the question under which circumstances rightsholders would be willing to join such distribution platforms in the first place. Whether rightsholders are better off inside or outside the bundle varies substantially across different revenue-sharing agreements. Evidence shows that—at least on average—on-demand streaming benefits rightsholders compared to the alternatives of transaction-based business models or online piracy.
Aggregate data indicates that unlicensed music consumption has continuously decreased since 2010, while consumption of unlicensed movie and TV show content has stagnated and has been slower to decrease. The underlying reason may lie in differences in the release strategies across industries, where music is made available immediately on all distribution channels, and movies are first released in physical channels and only later through online media.
Copyright in the age of AI
The most striking innovation developments in recent years regarding content creation have occurred with generative artificial intelligence (AI). As AI-generated output is reaching a point where it is difficult to distinguish it from works created by humans, the importance of human influence on creative works is put to the test, raising questions about originality and ownership. Currently, most jurisdictions do not protect AI-generated works because of the missing element of “human authorship,” which is seen as the guiding principle in copyright law, per the Berne Convention for the Protection of Literary and Artistic Works. Since these models are often trained on vast amounts of data available on the internet, often copyrighted, accessed through web scraping and Text and Data Mining (TDM) but without license agreements, questions arise whether this practice falls under copyright infringement or an exception of copyright still needs to be answered. The same goes for potential remuneration systems for rightsholders whose works are part of the training data.
Even in a de facto weakened copyright system, we have seen an increase in creative output as technological advancements and online platforms have simultaneously reduced entry barriers. Although academic research has made significant progress in advancing the understanding of the fundamental trade-offs of copyright law, its scope, and length, the setting for most empirical work on copyright focuses on the industries of books, music, and film. Only a few empirical studies examine software, databases, performance art, and photography. Since these areas are treated differently in copyright law, e.g., concerning a shorter protection period, the external validity of the existing evidence base might be diminished.
The socio-demographic diversity of creators and copyright holders, such as in terms of age and gender, is another vital yet understudied topic in the copyright literature. An empirical study shows that women published a third as many books as men in 1970. By 2020, women produced the majority of books, creating welfare gains for a wide range of consumers. Nevertheless, a report by the U.S. Copyright Office on copyright registrations from 1978 to 2020 reveals that in other areas of copyright, such as music and film, women continued to lag behind men in copyright registrations. It is important to understand these differences in order to develop and apply copyright law in ways that advance equality and inclusivity and promote participation in creative practices.
More research is necessary to better understand the role of the relationship between creators and intermediaries since there is variation across content types and platforms. With market-enabling solutions, such as technology that enables automated enforcement and algorithmic licensing, the question is whether the traditional institutions in the copyright system will or should become obsolete. Studying the relative benefits and costs of the copyright system and technology-based mechanisms is crucial, particularly regarding the welfare effects of derivative works under traditional and algorithmic licensing. Without an updated copyright system to reflect how new technologies, particularly AI, alter creators’ and users’ interests, the welfare of both creators and consumers may suffer.
Articles represent the opinions of their writers, not necessarily those of ProMarket, the University of Chicago, the Booth School of Business, or its faculty.