4 July 2024

Special Issue: AI and Law

Dear Readers

Artificial intelligence (AI) is currently on everyone's lips. The development and use of AI systems, which are affecting ever more areas of life, can only be properly understood from an interdisciplinary perspective. Technological understanding is a prerequisite, but politics and law determine the effective handling. Internationally, regulations are springing up like mushrooms; following the European Union with the Artificial Intelligence Act, the Council of Europe published a Framework Convention and the OECD revised the existing AI guidelines. At the end of 2023, the Swiss Federal Council announced its intention to draw up an overview of the legal issues arising by December 2024. However, this rather leisurely pace need not be a disadvantage, as Switzerland can incorporate foreign experience into its deliberations; moreover, there is no need to enact a comprehensive Swiss AI law (of over 100 pages), but the need for action can certainly be met with some specific (mainly vertical, sometimes also horizontal) legislative amendments and additions in Switzerland.

This focus issue of Jusletter IT takes up the breadth of the new legal issues from various perspectives: on the one hand, not all, but the important topics in the AI context are addressed; on the other hand, academic treatises can be found alongside more practice-oriented reports. The international regulations (Council of Europe, EU [with special attention to risk management], OECD) and topic-specific considerations (ensuring compliance with intellectual property rights, the problem of potential discrimination, guaranteeing the basic principles of data protection, structuring an appropriate liability regime) are discussed. The use of AI is also becoming increasingly important in the area of public law, which is why an explanation of the challenges for administrations and courts is essential.

This Jusletter IT is an interim step; the published descriptions and opinions are not set in stone, but form the starting point for further considerations within and outside the Weblaw community.

Rolf H. Weber
University of Zurich and Bratschi Attorneys at Law

Beiträge

Die KI-Konvention des Europarats: Ursprung, Inhalt, Ausblick

David Marti

The article explains the genesis, content and upcoming challenges of the Council of Europe Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law, which was adopted in May 2024 and represents the first binding international set of rules for AI systems. The author emphasizes the need to hold technology companies accountable at the beginning of the AI value chain and points to unresolved technical challenges to effectively implement the principles set out in the Convention about activities within the life cycle of AI systems.

Risikomanagement nach der EU-Verordnung über Künstliche Intelligenz

Martina Arioli

The risk management system is one of the main requirements for high-risk AI systems under the EU Act on Artificial Intelligence (AI Act). The AI Act follows the EU concept of the New Legislative Framework (NLF) for product safety. Basic requirements - such as requirements for an AI risk management system - are defined in the AI Act, whilst the technical implementation of these requirements shall be ensured by "harmonised standards". This article shall provide an overview of the requirements for the risk management system in accordance with the AI Act as well as existing standards for an AI risk management system.

Neue OECD-Instrumente zu künstlicher Intelligenz

Christine Kaufmann

This article provides an initial summary of recent, yet little known developments at the OECD to help governments and businesses promote trustworthy AI. The author summarises the current discussions and places them in the context of the OECD's leading instruments on due diligence. She concludes that the ongoing work on a due diligence guidance for AI has the potential to address the increasing fragmentation of AI-related standards in a particular area.

Diskriminierung beim Einsatz von Künstlicher Intelligenz (KI)

Florent Thouvenin

Stephanie Volz

Soraya Weiner

Christoph Heitz

The causes of discrimination by AI systems are diverse. They can lie in the definition of the model structure and training procedures and in non-representative training data, but also in the application of the systems. The risk of discrimination by AI systems can only be reduced through a combination of legal and technical measures. This requires a close cooperation between computer scientists and lawyers and a mutual understanding of terms and concepts. This article takes a first step by providing lawyers with the technical basics required to understand the emergence of discrimination by AI systems.

Künstliche Intelligenz und Datenschutz

Rolf H. Weber

In recent years, refined laws in most countries have sought to strengthen the basic principles of personal data protection. However, the latest technological developments, especially "artificial intelligence" (AI), are now leading to further (additional) challenges. The existing regulatory environment, properly interpreted and applied, is able to overcome most of the data protection problems caused by AI systems.

KI und Haftung: Lösungsansätze für die Schweiz

Isabelle Wildhaber

To have a modern Product Safety Act and a modern Product Liability Act, it is necessary to make revision to these two Acts. There is agreement on this necessity in Swiss academic literature. However, this pillar of AI liability has hardly been discussed in Swiss politics and the press to date. This should become a regulatory priority. A look at EU law shows: Compared to the European directives on product safety and on product liability (which are currently being revised), the practical impact of the proposed directive on AI liability appears to be less poignant. For Switzerland, I would not recommend to implement the proposed EU directive on AI liability nor would I recommend the introduction of strict AI liability. Instead, AI regulation should take up sector-specific regulations, especially as many AI damages will be covered by sectoral strict liability under national law, such as self-driving cars by traffic regulations or drones by aviation regulations.

KI-Projekte in der gerichtlichen Praxis

Frank Richter

The introduction of artificial intelligence (AI) in judicial practice has gained significance in recent years, particularly through projects like FRAUKE and JANO. These projects aim to optimize the processing of mass litigation and efficiently anonymize judicial decisions. FRAUKE assists judges in drafting judgments using text modules, while JANO pseudonymizes judicial decisions to make them suitable for publication. Both projects have been successfully piloted and are intended for nationwide implementation. The acceptance of these technologies among the judiciary is high, and they offer great potential for the future of the justice system.

KI in der Verwaltung: Entwicklungen und Herausforderungen

Nadja Braun Binder

Nina Laukenmann

Liliane Obrecht

This article focuses on the legal aspects of the use of artificial intelligence (AI) in public administration. Specific examples are used to illustrate possible areas of application and various legal challenges. In addition to the legislative developments in Switzerland, reference is also made to international regulatory efforts. The authors conclude that in Switzerland, too, at least some legal foundations are useful and necessary in order to meet the challenges and organize the use of AI in accordance with the rule of law.

Künstliche Intelligenz und Immaterialgüterrecht

Mathis Berger

This article provides an overview of the challenges that artificial intelligence poses to intellectual property law. The factual challenges and their legal answers are in flux, which is why an attempt is made to find solutions based on the principles of intellectual property law, which have led to partially similar regulations in intellectual property law.