Law (Copyright) vs. AI Art: Challenges from Data Input to Output Protection and Policy Recommendations for Thailand

Abstract

This academic article aims to explore the gaps in copyright law caused by generative AI art by assessing its challenges to the Copyright Act B.E. 2537 (1994) in two dimensions. The study reveals limitations in the current legal framework: (1) utilizing copyrighted works for text and data mining (TDM) in AI training (input) creates legal uncertainty when analyzed through the fair dealing mechanism under Section 32; and (2 )
the definition of “author” under Section 4 and the Supreme Court’s jurisprudence based on the “sweat of the brow” doctrine fail to provide
clarity on whether the inputting of prompts entails sufficient effort to warrant copyright protection for the output. A comparative analysis of the approaches in the United States, the
European Union, the United Kingdom, and the People’s Republic of China demonstrates a global trend toward striking a balance between protecting artists and promoting innovation. Consequently, the author proposes
three policy recommendations for Thailand: (1 ) enacting a specific copyright exception for text and data mining (TDM) coupled with an opt-
out mechanism; (2 ) issuing clear guidelines by the Department of Intellectual Property to establish criteria for human contribution; and (3) creating transparency and benefit-sharing mechanisms. These recommend dations should be integrated into the ongoing consideration of the draft amendment to the Copyright Act to establish legal norms that are well-
suited for the digital economy.