布鲁盖尔研究所-支持减少对生成人工智能输入和输出的版权保护的经济论据（英）-2024.4-23页_268kb

Economic arguments in favour of reducing copyright protection for generative AI inputs and outputs Bertin Martens argues that copyright on GenAI training inputs is economically inefficient and leads to overprotection. He suggests making text and data mining (TDM) exceptions unconditional to promote innovation and competition. For GenAI outputs, the marginal cost is nearly zero, reducing incentives for piracy and making exclusive copyright unnecessary. Legal barriers like creator moral rights hinder proper economic analysis and might lead to market failures. Martens calls for broadening TDM exceptions or weakening opt-out conditions to foster innovation and prevent stifling economic growth.