Search

Several emerging technologies afford a specific type of interference with privacy which has not yet been conceptualised: inferring information about mental properties of persons on the basis of physiological or biometric signals, in short: mind probing. This paper conceptualises it, analyses several forms and proposes concrete norms that ban and criminalise it. Facial emotion recognition, neuroimaging, or wearable consumer devices such as smartwatches process biometric data to draw inferences about minds; the prevalence and sophistication of such mind probes will soon increase through AI data analysis. This paper explores how human rights law, the EU General Data Protection Regulation and the EU AI Act address mind probes and shows why the current level of protection appears insufficient. This motivates novel norms banning and criminalising non-consensual biometric-based forms of mind reading, ensuring that at least some parts of the mind remain in principle free from non-consensual mind probing. [ABSTRACT FROM AUTHOR]

The lack of explainability of Artificial Intelligence (AI) is one of the first obstacles that the industry and regulators must overcome to mitigate the risks associated with the technology. The need for 'eXplainable AI' (XAI) is evident in fields where accountability, ethics and fairness are critical, such as healthcare, credit scoring, policing and the criminal justice system. At the EU level, the notion of explainability is one of the fundamental principles that underpin the AI Act, though the exact XAI techniques and requirements are still to be determined and tested in practice. This paper explores various approaches and techniques that promise to advance XAI, as well as the challenges of implementing the principle of explainability in AI governance and policies. Finally, the paper examines the integration of XAI into EU law, emphasising the issues of standard setting, oversight, and enforcement. [ABSTRACT FROM AUTHOR]

Rapidly emerging generative artificial intelligence (GenAI) models stand at the epicentre of current public discourse. They demonstrate impressive abilities to generate various types of data promptly and cost-effectively. However, AI developers need to train their systems on massive volumes of data which is usually copyrighted. Therefore, the growth of copyright-related concerns in the field of GenAI comes as no surprise. The study introduces two solutions which could mitigate the tension between copyright holders and AI developers, one legal (text and data mining (TDM) exceptions of the CDSM Directive) and one technical (synthetic data), highlighting the promises and challenges of both. First, the article will discuss the capability of TDM exceptions to facilitate the fundamental right to information and the freedom of research in the context of AI development. Next, the paper will analyse how providers of GenAI models can leverage synthetic data to comply with copyright law while training their systems and what risks might be associated with this approach. The findings of this study will indicate what issues, in both legal and technical spheres, should be addressed to ensure a balance of powers in the digital environment and effective functionality of the EU AI sector. [ABSTRACT FROM AUTHOR]

This paper explores the regulatory sandbox regime under the EU's draft Artificial Intelligence (AI) Act. It investigates how useful the sandbox regime is for testing an AI-based skin cancer detection systems in an EU member state. The paper focuses on whether the proposed AI regulatory sandbox regime can resolve tensions between innovation and safety. Although we find considerable potential for the sandbox regime, the proposal also creates several legal issues. It blurs jurisdictional boundaries between the EU and member states, raises concerns of legality and equal treatment, creates liability risks for innovators, and fails to require informed consent from testing subjects. To address these problems, the paper suggests adopting a more targeted legal basis for the sandbox regime that takes inspiration from conventional testing mechanisms such as clinical investigations for medical devices. [ABSTRACT FROM AUTHOR]

This paper asks how lawmakers and other stakeholders envision the potential benefits and challenges arising from Artificial Intelligence (AI). A close reading of the European Union's draft AI Act, a bill proposed by the European Commission in April 2021, and of 302 response papers submitted by NGOs, businesses and business associations, trade unions, academics, public authorities, and citizens, shows that pluralistic sociotechnical imaginaries contest: (1) the essential characteristics of technology as they relate to society, politics, and law; (2) whether, how and how much law can enable, direct or constrain scientific & technological developments; and (3) the degree to which law does and should intervene into scientific & technological controversies. The feedback from stakeholders reveals major disagreements with the lawmakers in terms of how the relevant characteristics of AI should influence legal regulation, what the desired law should look like, and whether and how the law should intervene into expert debates in AI. What is more, different types of stakeholders diverge considerably in what they problematise and how they do so. [ABSTRACT FROM AUTHOR]

The paper argues that the material scope of AI regulations should not rely on the term 'artificial intelligence (AI)'. The argument is developed by proposing a number of requirements for legal definitions, surveying existing AI definitions, and then discussing the extent to which they meet the proposed requirements. It is shown that existing definitions of AI do not meet the most important requirements for legal definitions. Next, the paper argues that a risk-based approach would be preferable. Rather than using the term AI, policy makers should focus on the specific risks they want to reduce. It is shown that the requirements for legal definitions can be better met by defining the main sources of relevant risks: certain technical approaches (e.g. reinforcement learning), applications (e.g. facial recognition), and capabilities (e.g. the ability to physically interact with the environment). Finally, the paper discusses the extent to which this approach can also be applied to more advanced AI systems. [ABSTRACT FROM AUTHOR]

Artificially intelligent machines are increasingly capable of accomplishing tasks that have until now been considered exclusively human abilities. One such task is the ability to invent. In 2021, an Australian court became the first court in the world to recognise an Artificially Intelligent system as the inventor in a patent application, raising the question of whether AI systems could be considered the responsible actor of their actions, instead of the human using the AI. This paper examines the question of autonomy: how the concept of autonomy underpins the granting of legal rights to humans, and whether it is justified to grant legal rights to AI systems that are autonomous. We propose that there are two distinct and separate concepts of autonomy, Person Autonomy and Machine Autonomy, and that this distinction can guide how the law should treat AI systems. [ABSTRACT FROM AUTHOR]

Despite their importance in the ethics of digital technologies, trust and trustworthiness do not appear in the international law on digital technologies. This surprising absence is the point of departure of this paper. On the basis of a survey of philosophical and empirical scholarship on trust in the context of digital technologies (especially AI), the article proceeds by identifying relevant norms in international law which can be considered to operationalise relevant dimensions of trust and trustworthiness. By clustering these norms, a technical conception of trust appears, focused on securing the trustworthiness of an AI system. This conception is based on the framing of an AI system as an agent and on the premise that the role of the law is to promote the uptake of technologies. This technical conception of trust should be enriched by way of three relevant extensions, each addressing in turn: the responsibility of developers, the power relations between users and developers of AI systems, and the benefits associated with the use of AI systems. This extended conception of trust also extends the realm of relevant legal norms that can be used to address AI-powered actions. [ABSTRACT FROM AUTHOR]

The ubiquitous use of artificial intelligence (AI) can bring about both positive and negative consequences for individuals and societies. On the negative side, there is a concern not only about the impact of AI on first-order discrete individual rights (such as the right to privacy, non-discrimination and freedom of opinion and expression) but also about whether the human rights framework is fit for purpose relative to second-order challenges that cannot be effectively addressed by discrete legal rights focused upon the individual. The purpose of this article is to map the contours and utility of the concept of human dignity in addressing the second-order challenges presented by AI. Four key interpretations of human dignity are identified, namely: non-instrumentalization of the human person; the protection of certain vulnerable classes of persons; the recognition and exercise of inherent self-worth (including the exercise of individual autonomy); and a wider notion of the protection of humanity. Applying these interpretations to AI affordances, the paper argues that human dignity should foreground three second-order challenges, namely: the disembodiment of empiric self-representation and contextual sense-making; the choice architectures for the exercise of cognitive autonomy; and, the experiential context of lived experiences using the normative framework of human vulnerability. [ABSTRACT FROM AUTHOR]

While artificial intelligence technologies (AI), such as machine learning (ML), hold significant potential for the economy and social wellbeing, it is unclear to what extent copyright laws stimulate or impede the development of these promising technologies. The unauthorised use of copyright-protected works in the ML process and its possible implications on economic rights of authors have been previously explored, however, the implications of such use on the moral rights of authors – the rights of attribution and integrity – have not been examined. This paper, by focusing on Australia as a case study, explores whether the use of works as training data in the ML process could amount to the infringement of moral rights of authors and, if so, whether law reform in the area is needed. [ABSTRACT FROM AUTHOR]

When combined with an appropriate level of human judgement, machine learning applications were crucial resources insupporting decision-making in the context of the Covid-19 crisis, resulting in more efficient and better-informed responses to ethicalissues. This paper focusses on four social dimensions (bioethical, political, psychological, and economic) from which the decisionstaken in the context of the Covid-19 crisis derived major ethical implications. On the one hand, I argue against the possibility ofaddressing these issues from a purely algorithmic approach, elaborating on two types of limitations for automated systems toaddress ethical issues. This leads me to discuss how different ethical situations call for different performance metrics with regards tothe 'contextual explicability and performance issue', as well as to enunciate a gold principle: 'legitimacy trumps accuracy'. On the otherhand, I present practical examples of machine learning applications which enhance, instead of dilute, human moral agency in betteraddressing these issues. I also suggest a 'moral perimeter' framework to ensure the responsibility of algorithms-assisted decisionmakersfor critical decisions. The unique potential of AI to 'solve' moral dilemmas by intervening on their conditions of possibility thenprompts me to discuss a new type of moral situation: AI-generated meta-dilemmas. [ABSTRACT FROM AUTHOR]