Copyright on the internet: achieving security through electronic devices an artificial intelligence approach

Abstract

This thesis aims to provide a novel approach to ensure copyright compliance online, appropriate for the Internet of Things and the robotic revolution. To achieve this, three different aims are pursued: - A novel application of “by design” solutions to copyright protection is introduced and its advantages and disadvantages discussed from a jurisprudential and doctrinal perspective.

- On the basis of this, a new theoretical framework for legal AI is developed that draws on Andy Clark’s “extended mind” model of cognition. - This then leads to a new way for implementing legal knowledge and cognition in expert systems, and an outline implementation of this approach is introduced and discussed.

The Internet and the Information and Communication Technology (ICT) revolution has delivered important social benefits that had reshaped our conception of the world by expanding knowledge and culture. However, its dynamic nature and technical composition had made it attractive for the commission of illegal activities also. One of the areas that have experienced this combined effect is the creative industry.

The Internet has enhanced its capacity to reach new markets, distribute their material digitally, has allowed aspiring artists to disseminate their work to large audiences and has even incentivised the development of novel techniques to create new material. Nevertheless, it has also allowed unauthorized access and distribution of copyrighted materials, causing economic harms on a considerable scale.

Traditional legal responses -punishment after a criminal act- proved incapable to provide a deterrent effect to these activities, mostly due to the capacity of infringers to adapt to the operational particularities of cyberspace. The difficulty to enforce copyright in cyberspace, outside national jurisdictions and with easier and faster ways of replication of material, led to the development of a new paradigm of enforcement: automated enforcement by software code. Here, automated enforcement is delivered by embedding copyright law directly into software code: Digital Rights Management, commonly known as DRM was the first adopter of this approach. These programs provided restricted access, based solely on the rights granted by the author or the rights holder in relation to a particular work. This, arguably, had the potential of reducing the possibilities of unlawful use. But it also prevented some forms of legal use, especially those exceptions to copyright granted by statute. Despite the positive expectations around DRM, it therefore did not have the expect success. The thesis will discuss some of the lessons that can be learned from this relative failure. One result will be a discussion if these systems were “too dumb” to present truly regulation by code, and if this contributed to their failure.

However, the failure of these early devices does not necessarily mean the end of the collaboration between copyright law and technology. The concept of “law as computer code” was addressed from other technological approaches suitable of operating without the issues presented on DRM.

The thesis argues therefore that robotic technology is a novel application for the implementation of automated enforcement of copyright law. In this scenario, artificial creators require the capacity to understand and replicate human cognitive processes used during the creation of artistic works. This includes the detection and lawful management of relevant features contained in already existing works. To avoid this situation, these devices need the capacity to adapt their behaviour based on the legal status of a relevant work to reduce the risk of performing copyright infringing conducts. This represents a significant advance in relation to traditional devices: legal compliance is delivered through a preventive, non-punishing scheme, which presents an effective method to protect digital material on the cyberspace.

Additionally, this research addresses, but to a lesser degree, the impact of computer-generated works on the legal framework. It is a very well established principle that copyright protects human creativity and spontaneity, and one can ask if such characteristics are attributable to a computer. This has left these creations outside of the protective scope of most legal systems. This also means the development of an international approach is a difficult task. This research will demonstrate how legal ontologies can help to address this problem.

Among the novel aspect of the thesis, the combination of computational legal reasoning and theory of legal cognition stands out. In this sense, traditional implementations of legal artificial intelligence aimed to replicate the reasoning processes performed by a judge as part of a trial or similar formal legal procedures. This means a high degree of legal accuracy was required, which in turn necessitated a considerable amount of time and resources to develop a formal reasoner and knowledge base.

Contrary to this, cognitive operations performed by ordinary humans are rarely a straightforward process, they require the capacity to interact with the environment and leave little time for complex reasoning processes. Instead, citizens implement a series of cognitive strategies that involve interaction with the environment, cognitive shortcuts and the implementation of preconceived responses. This can be seen in a variety of scenarios, for example, in the presence of an image of a camera crossed by a diagonal line indicates to us that the action of taking photographs is not allowed and it probably carries a legal consequence; a closed door signals a prima facie prohibition to enter that space without a key etc. These actions allow us to operate law compliantly in a dynamic social environment without performing unnecessary complex legal processes or the constant conseil of an attorney. The research presented in this thesis, aims to define if this intuitive reasoning about the law in open scenarios, can be implemented in computational legal systems. Finally, this thesis concludes that by delivering intelligent technology the capacity to replicate legal cognitive processes, automated enforcement of copyright law can be achieved.