Position Paper: A Legal Framework for Artificial Intelligence

AlgorithmWatch Switzerland co-authored a position paper on the regulation of AI systems in Switzerland. The paper was written in collaboration with an interdisciplinary consortium of academics and supported by the DSI Strategy Lab (University of Zurich).
Maxime Valcarce | Unsplash

Today, algorithmic Decision Making-Systems (ADM systems), often known and discussed as «Artificial Intelligence», are everywhere – they are used, e.g., to evaluate creditworthiness, to predict detainee’s risk of recidivism, or to target ads on Social Media. While the deployment of these systems promises great benefits, it often comes with risks and challenges for both individuals and society.

This has implications at the legal level: A legal system must address these risks and challenges. While the EU has already proposed a regulation for AI-systems (“Draft Artificial Intelligence Act”) and while the Council of Europe is preparing a legal framework on AI, in Switzerland, the debate is now being launched.

In collaboration with academics of various disciplines, Anna Maetzener and Matthias Spielkamp (representing AlgorithmWatch Switzerland) have co-authored the position paper “A Legal Framework for Artificial Intelligence”.

The paper does not propose a comprehensive horizontal regulation of ADM systems but a combination of general and sectoral norms. It requires a detailed analysis, identifying legal gaps and ways to close them, and considering existing laws that need to be reinterpreted or reapplied. It focuses on five areas:

  • Transparency, Recognizability and Comprehensibility
  • Discrimination
  • Manipulation
  • Liability
  • Data protection and security

For example, to promote transparency, the paper suggests an obligation to label ADM systems or to establish public registers on ADM systems deployed in the public sector. A general equal treatment law could be an important tool to address discrimination among private parties. And for some deployments, there seems to be no alternative but to introduce bans or moratoria: The use of facial or other kinds of biometric recognition systems in public space, which establishes mass surveillance, is not compatible with fundamental rights. The same goes for Social Scoring systems.

Read our position paper  «A Legal Framework for Artificial Intelligence[AM1] ». 

The position paper was presented and discussed at a public event in November 2021, organized by the DSI Strategy Lab, the Center for Information, Technology, Society (ITSL) at the University of Zurich, and the University of Basel.

The videos of the presentations are available at the DSI Strategy Lab website.

The panel discussion with Angela Mueller (AlgorithmWatch Switzerland) can be streamed here [in German]: