This country-specific Q&A provides an overview of artificial intelligence laws and regulations applicable in Switzerland.
What is the legal definition of artificial intelligence in Switzerland?
A universally accepted description of artificial intelligence (AI) remains elusive, as existing definitions often revolve around human intelligence rather than machine-based intelligence. These formulations pose challenges by necessitating the definition of "intelligence," particularly the human variety.
Given this complexity, the Swiss government has not yet established a comprehensive AI definition that applies universally across all domains.
In a report titled "Artificial intelligence and International Rules" dated April 13, 2022, the Federal Department of Foreign Affairs (FADF) of Switzerland provides the following characterization of AI: "Artificial intelligence (AI), also referred to as 'machine learning,' involves the creation or programming of computers to perform tasks typically requiring human or biological intelligence. These tasks encompass functions such as visual perception (image recognition), speech recognition, language translation, visual interpretation, and rule-based game playing. AI pertains to the development of intelligent machines capable of executing activities that humans ordinarily engage in, signifying machines with self-learning capabilities that exhibit 'intelligent' behaviour.'"
Are there any civil and criminal liability rules in Switzerland that may apply in case of damage caused by artificial intelligence systems?
The legal complexities surrounding AI-induced harms are illuminated through tort law and product liability, revealing challenges within existing frameworks. These challenges stem from AI-specific attributes, such as autonomy and service-based nature, complicating the assignment of responsibility among multiple stakeholders. This intricacy is addressed through three civil liability frameworks: tort liability, where Swiss law necessitates compensation for unlawful damage, contingent on four criteria; product liability, holding producers accountable for product defects and extending to AI components in specific cases; and contractual liability, which shifts the burden of proof based on contractual contexts, potentially leading to gaps in AI-related liability.
Amid evolving technology, the risk-oriented approach in civil liability emphasizes the accountability of technology beneficiaries. However, criminal liability for AI remains uncertain within Swiss law, with current provisions lacking specific guidance for AI-related criminal behaviour, underscoring the limitations tied to AI's legal standing and culpability. In essence, the multifaceted landscape of AI-induced harms demands nuanced adaptations within these liability paradigms to address the unique attributes and consequences of artificial intelligence.
Who is responsible for any harm caused by an artificial intelligence system in Switzerland?
The answer depends on which liability regime is triggered.
The issue of accountability is gaining prominence as artificial intelligence gains greater autonomy. However, from a legal perspective, artificial intelligence is classified as an "entity," not an "individual." Despite its increasing autonomy, Swiss civil law does not confer legal personality upon artificial intelligence; only natural or legal persons can be held accountable in cases where harm arises from an AI system . This principle remains relevant even when artificial intelligence operates without direct human supervision . According to Swiss legal principles, an AI system is deemed incapable of discernment and lacks the capacity for intentional or negligent actions . Consequently, no culpability can be attributed to an AI system.
- Conseil fédéral, Défis de l’intelligence artificielle, Rapport du groupe de travail interdépartemental « Intelligence artificielle » au Conseil fédéral, 2019, p. 36ss ; Yaniv Benhamou/Justine Ferland, op. cit, p. 165.
- Conseil fédéral, Défis de l’intelligence artificielle, Rapport du groupe de travail interdépartemental « Intelligence artificielle » au Conseil fédéral, 2019, p. 36ss.
For the full article, please read The Legal 500's -> Switzerland: Artificial Intelligence