Abstract

As the use of AI in the financial sector is ever growing, so are the legal challenges, both in amount, and in complexity. The use of AI in the sector of financial advicehas made the working process both easier and more effective for those offering financial advice to consumers, while the technology as such still is far from finished, or completely understood by those who work with it on a day to day basis, both as developers and as users. The focus in this thesis is how the field of law is cooperating with the field of data science to solve legal challenges arising when a financial advisor is giving advice to a consumer while using an AI-system.

In the financial sector, AI is used to make predictions and make decisions. These has to be as precise as possible to appease the consumers and give the financial advisor reason to actually use the AI-system to its benefit. To ensure that, the input that the developers of the AI-system gives to the AI-system has to be correctly sorted and marked by the system. The development of an AI-system is complex and can consist of both machine learning, deep learning, and artificial neural networks.

When a consumer claims that a financial advisor has been liable of causing the consumer financial damage, he or she has to prove that the financial advisor hasbeen guilty of acting carelessly. This is done by a comparison between the actions of the financial advisor and the requirements deriving from the Swedish Securities Markets Act and the recommendations that is developed by the European Securities and Markets Authority (ESMA). The legal investigation that has to be done by those who practice law in regards to challenges where AI is used contains complex questions that has to be asked in regard to the burden of proof, causality and adequacy.

The European Union has since Ursula von der Leyens accession as the President of the European Commission actively been working on several legal documents that aims to clarify which requirements developers and users of AI-systems has to adhere to, which also can be used to evaluate whether or not an operator of said system can be held liable due to careless usage of the AI-system. Even though several legal documents has been drafted, legal challenges still existin regard to the burden of proof, causality, and adequacy. The drafted documents may entail distinct legal alteration of how a challenge may be faced, but there are still challenges that will arise due to the fact that questions of adequacy and causality isn’t regulated in the written law.