Jusletter

Table of contents

• 1. Introduction
• 2. Primary data protection risks from AI
• 3. Protections available under Swiss law
• 3.1. Introduction
• 3.2. Under the nFADP
• 3.2.1. Mass surveillance
• 3.2.1.1. Biometric data
• 3.2.1.2. Data protection impact assessment
• 3.2.1.3. Log
• 3.2.1.4. Processing regulations
• 3.2.2. Automated processing
• 3.2.2.1. Profiling
• 3.2.2.2. Automated decisions
• 3.3. Other sources
• 3.3.1. International
• 3.3.1.1. Convention 108+
• 3.3.1.2. European Human Rights Convention
• 3.3.2. Domestic
• 3.3.2.1. Constitution
• 3.3.2.2. Criminal Procedure Code
• 3.3.2.3. Labour Law and Employment Law
• a. Labour Law
• b. Employment Law
• 4. Proposed AI-specific protections in the EU: the AI Act
• 4.1. Introduction to the AI Act
• 4.1.1. Background
• 4.1.2. Overview
• 4.2. Unacceptable risk
• 4.2.1. Scope
• 4.2.2. Rules
• 4.3. High risk
• 4.3.1. Scope
• 4.3.2. Rules
• 4.4. Limited risk
• 4.4.1. Scope
• 4.4.2. Rules
• 4.5. Minimal risk
• 4.5.1. Scope
• 4.5.2. Rules
• 4.6. General-purpose AI, foundation models and generative AI
• 4.6.1. Scope
• 4.6.2. Rules
• 4.7. Innovation
• 5. Comparison
• 5.1. Personal data protection
• 5.1.1. Unacceptable risk
• 5.1.2. High risk
• 5.1.3. Limited risk
• 5.1.4. Minimal risk
• 5.1.5. General-purpose AI/foundation models and generative AI
• 5.2. Advantages of the Swiss approach
• 5.2.1. Nuanced approach
• 5.2.1.1. Exclusion
• 5.2.1.2. Overreaching
• 5.2.2. Innovation
• 5.3. Disadvantages of the Swiss approach
• 5.3.1. Piece-meal solution
• 5.3.1.1. Lack of legal certainty
• 5.3.1.2. Unaddressed risks
• 5.3.2. Timing
• 6. Conclusion