Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

AI och ansvarsfördelning vid skador - Ansvar utan gränser? Rättsliga utmaningar i en autonom värld

Bahmany, Arina LU (2024) LAGF03 20242
Department of Law
Faculty of Law
Abstract
The rapid development of artificial intelligence (AI) has led to significant ad-vancements but also to new legal challenges. Technologies such as autono-mous vehicles and advanced AI systems have raised questions about how responsibility for damages caused by AI systems should be allocated. This is particularly problematic given that traditional product liability principles were not designed to address the adaptive and autonomous characteristics of AI.
To address these challenges, the EU has proposed new directives and regula-tions aimed to modernizing the legal framework and tackling issues related to liability. This thesis examines how damages resulting from AI systems impact the legal allocation of responsibility and analyzes whether... (More)
The rapid development of artificial intelligence (AI) has led to significant ad-vancements but also to new legal challenges. Technologies such as autono-mous vehicles and advanced AI systems have raised questions about how responsibility for damages caused by AI systems should be allocated. This is particularly problematic given that traditional product liability principles were not designed to address the adaptive and autonomous characteristics of AI.
To address these challenges, the EU has proposed new directives and regula-tions aimed to modernizing the legal framework and tackling issues related to liability. This thesis examines how damages resulting from AI systems impact the legal allocation of responsibility and analyzes whether the proposed re-forms can be considered adequate solutions to these problems.
AI systems are often complex and the result of collaboration between multiple actors, making it difficult to clearly identify who is responsible for damages. A central theme of this thesis is defining how liability should be allocated when multiple parties are involved, such as software developers, hardware manufacturers, and users. The thesis also highlights how traditional concepts like "product" and "defect" need to be adapted to include both physical and digital components of AI systems. A defect may arise from hardware, algo-rithms, or data management, and this diversity of potential fault sources in-creases the complexity of the legal landscape.
Furthermore, the reforms have a great potential to address the shifting burden of proof faced by injured parties in AI-related cases. The technical complexity of AI systems and the fact that they often learn and evolve after integration on the market makes it difficult for injured parties to demonstrate that a specific defect directly caused the harm. The proposed reforms, which include clearer requirements for documentation and transparency, aim to simplify this process and enhance legal certainty.
The thesis concludes by stating that the proposed reforms represent an im-portant step forward in addressing the legal challenges presented by AI tech-nology. By modernizing definitions, responsibility allocation and rules regard-ing the burden of proof, the ability to manage damages caused by AI systems should be improved. At the same time, there remains a need for further development to ensure that the regulation is both fair and adapted to the technologi-cal reality. Balancing the protection of consumers with the promotion of continued innovation in AI is crucial, and this is where the EU's work on product liability legislation can play a decisive role. (Less)
Abstract (Swedish)
Den snabba framväxten av artificiell intelligens (AI) har lett till betydande framsteg, men också till nya juridiska utmaningar. Teknologier som autonoma fordon och avancerade AI-system har aktualiserat frågor kring hur ansvar för skador som orsakas av AI-system ska fördelas. Detta är särskilt problematiskt i ljuset av den traditionella produktansvarsprinciper som inte är utformad för att hantera AI:s adaptiva och autonoma egenskaper.
För att möta dessa utmaningar har EU föreslagit nya direktiv och förordning-ar, som syftar till att modernisera regelverket och hantera ansvarsproblemati-ken. Denna uppsats undersöker hur skador som uppstår till följd av AI-system påverkar den rättsliga ansvarsfördelningen och analyserar om de före-slagna... (More)
Den snabba framväxten av artificiell intelligens (AI) har lett till betydande framsteg, men också till nya juridiska utmaningar. Teknologier som autonoma fordon och avancerade AI-system har aktualiserat frågor kring hur ansvar för skador som orsakas av AI-system ska fördelas. Detta är särskilt problematiskt i ljuset av den traditionella produktansvarsprinciper som inte är utformad för att hantera AI:s adaptiva och autonoma egenskaper.
För att möta dessa utmaningar har EU föreslagit nya direktiv och förordning-ar, som syftar till att modernisera regelverket och hantera ansvarsproblemati-ken. Denna uppsats undersöker hur skador som uppstår till följd av AI-system påverkar den rättsliga ansvarsfördelningen och analyserar om de före-slagna reformerna kan betraktas som adekvata lösningar på dessa problem.
AI-system är ofta komplexa och ett resultat av samarbete mellan olika aktörer, vilket gör det svårt att tydligt identifiera vem som bär ansvar vid skador. Ett centralt tema i uppsatsen är att definiera hur ansvarsfrågan bör hanteras när flera aktörer är inblandade, exempelvis utvecklare av mjukvara, tillverkare av hårdvara och användare. Uppsatsen lyfter också fram hur traditionella begrepp som "produkt” och ”säkerhetsbrist" behöver anpassas för att omfatta både fysiska och digitala komponenter i AI-system. En säkerhetsbrist kan uppstå i allt från hårdvara till algoritmer och datahantering, och denna mångfald av möjliga felkällor gör det juridiska fältet mer komplext.
Vidare diskuteras reformernas potential att hantera den förändrade bevisbördan som skadelidande möter i fall som rör AI. Den tekniska komplexiteten och det faktum att AI-system ofta lär sig och utvecklas efter lansering gör det svårt för skadelidande att visa att en viss säkerhetsbrist direkt orsakat en skada. Reformförslagen, som inkluderar tydligare krav på dokumentation och transparens, syftar till att underlätta denna process och förbättra rättssäkerhet-en.
Uppsatsen avslutas med en slutsats där de föreslagna reformerna utgör ett viktigt steg framåt för att möta de rättsliga utmaningar som AI-teknologin skapar. Genom att modernisera definitioner, ansvarsfördelning och regler kring bevisbörda förbättras möjligheterna att hantera skador som orsakas av AI-system. Samtidigt kvarstår behovet av ytterligare utveckling för att säkerställa att regleringen är både rättvis och anpassad till den teknologiska verkligheten. Balansen mellan att skydda konsumenter och främja fortsatt innovation inom AI är avgörande, och det är här EU:s arbete med produktansvarslagstiftningen kan spela en avgörande roll. (Less)
Please use this url to cite or link to this publication:
author
Bahmany, Arina LU
supervisor
organization
course
LAGF03 20242
year
type
M2 - Bachelor Degree
subject
keywords
Ai, produktansvar, skadeståndsrätt
language
Swedish
id
9180238
date added to LUP
2025-03-20 13:51:19
date last changed
2025-03-20 13:51:19
@misc{9180238,
  abstract     = {{The rapid development of artificial intelligence (AI) has led to significant ad-vancements but also to new legal challenges. Technologies such as autono-mous vehicles and advanced AI systems have raised questions about how responsibility for damages caused by AI systems should be allocated. This is particularly problematic given that traditional product liability principles were not designed to address the adaptive and autonomous characteristics of AI.
To address these challenges, the EU has proposed new directives and regula-tions aimed to modernizing the legal framework and tackling issues related to liability. This thesis examines how damages resulting from AI systems impact the legal allocation of responsibility and analyzes whether the proposed re-forms can be considered adequate solutions to these problems.
AI systems are often complex and the result of collaboration between multiple actors, making it difficult to clearly identify who is responsible for damages. A central theme of this thesis is defining how liability should be allocated when multiple parties are involved, such as software developers, hardware manufacturers, and users. The thesis also highlights how traditional concepts like "product" and "defect" need to be adapted to include both physical and digital components of AI systems. A defect may arise from hardware, algo-rithms, or data management, and this diversity of potential fault sources in-creases the complexity of the legal landscape.
Furthermore, the reforms have a great potential to address the shifting burden of proof faced by injured parties in AI-related cases. The technical complexity of AI systems and the fact that they often learn and evolve after integration on the market makes it difficult for injured parties to demonstrate that a specific defect directly caused the harm. The proposed reforms, which include clearer requirements for documentation and transparency, aim to simplify this process and enhance legal certainty.
The thesis concludes by stating that the proposed reforms represent an im-portant step forward in addressing the legal challenges presented by AI tech-nology. By modernizing definitions, responsibility allocation and rules regard-ing the burden of proof, the ability to manage damages caused by AI systems should be improved. At the same time, there remains a need for further development to ensure that the regulation is both fair and adapted to the technologi-cal reality. Balancing the protection of consumers with the promotion of continued innovation in AI is crucial, and this is where the EU's work on product liability legislation can play a decisive role.}},
  author       = {{Bahmany, Arina}},
  language     = {{swe}},
  note         = {{Student Paper}},
  title        = {{AI och ansvarsfördelning vid skador - Ansvar utan gränser? Rättsliga utmaningar i en autonom värld}},
  year         = {{2024}},
}