유럽연합 인공지능법(EU AI Act) 제정의 저작권법적 시사점

EU AI Act officially came into effect in August 2024 after the final approval of the European Council in May 2024, and is strengthening the transparency and accountability of AI systems through regulations that will be applied sequentially from 2025. The EU AI Act mainly includes a ban on the use of facial recognition technology and crime prediction tools by the police, mandatory disclosure of data sources, and stipulations on legal compliance of all AI systems. Violations of the above regulations may result in fines of up to 35 million euros or 7% of turnover. The EU AI Act is likely to become a global standard due to the Brussels effect, and its impact on the world is expected to be significant. In this paper, we analyzed the main contents of the EU AI Act, derived issues related to copyright law, and reviewed the implications and legal challenges. First, in terms of how the EU AI Act is linked to our country’s copyright law, it was mentioned that summary information on works used in AI system training should be disclosed in accordance with the transparency principle, and that text and data mining (TDM) regulations should be applied in accordance with the DSM directive, but compliance should be observed if the rights holder reserves use. Next, in order for our country to create a safe and reliable AI system environment and protect the legitimate rights holders of copyrighted works, it is urgent to establish legal provisions that can regulate AI system contents, and further, it was argued that AI providers and distributors should clearly recognize the scope of exemption and exceptions by introducing the TDM concept.