Back to Search Start Over

Infringement of intellectual property rights in the course of adversarial attacks on artificial intelligence systems from the perspective of European Union law

Authors :
Bohatchuk, Daria
Bohatchuk, Daria
Publication Year :
2024

Abstract

Artificial intelligence is having an increasing impact in many areas. At the same time, artificial intelligence systems can be targeted by socalled "adversarial attacks". Appropriate measures are needed to counter adversarial attacks on artificial intelligence, including with the help of law. The paper aims to analyze whether adversarial attacks on artificial intelligence systems constitute infringement of intellectual property rights to such systems or elements thereof. The paper focuses on copyright law, patent law, and sui generis database protection and examines how intellectual property rights can be infringed in the course of conducting adversarial attacks. It is concluded that the copying of artificial intelligence system or its parts and components (if they are copyrighted) in the course of an adversarial attack may qualify as reproduction and constitute copyright infringement if certain conditions are met. In this context, it is important to take into account that the act of reproduction committed for adversarial attack generally cannot be considered as accidental, but on the contrary, may be considered as intentional. When considering possible patent infringements in the context of adversarial attacks, in particular, the doctrine of equivalents and the principle of "exhaustion of rights" must be taken into account. The sui generis database right, which is an additional protection possibility for databases in the European Union, may also be infringed within the adversarial attacks under certain conditions. The need to include provisions on the unlawful character of adversarial attacks in the relevant legislation is determined. The provisions prohibiting adversarial attacks on artificial intelligence in the license agreements can also increase the level of legal protection of artificial intelligence systems.

Details

Database :
OAIster
Notes :
English
Publication Type :
Electronic Resource
Accession number :
edsoai.on1472946967
Document Type :
Electronic Resource