生成式AI被滥用如何治理？学者建议用好现有规则发展中规范

Core Viewpoint - The emergence of generative artificial intelligence (AI) raises significant legal challenges, particularly concerning copyright and infringement issues, necessitating a careful regulatory approach that utilizes existing laws rather than rushing into new legislation [2][3][4] Group 1: Current Legal Framework and Recommendations - Current specialized legislation for AI infringement is considered premature; instead, existing laws such as the Civil Code and Personal Information Protection Law should be leveraged to address AI-related infringement issues [3][4] - The approach should focus on "regulating in development," utilizing existing legal frameworks to interpret and apply rules effectively while accumulating case law and judicial interpretations [4][10] Group 2: Infringement Liability and Standards - The core legal issue in AI infringement revolves around the choice of liability principles, with a preference for "fault liability" over "strict liability" to avoid stifling AI development [5][6] - The standard for determining fault in AI infringement should center on "breach of duty of care," considering specific scenarios to balance risk management costs with reasonable obligations [6][7] Group 3: Deepfake Technology and Personal Information - There is a need to prohibit the use of deepfake technology to infringe on others' rights, with recommendations to interpret existing laws to address deepfake-related infringements effectively [8][9] - The treatment of publicly available personal information should differentiate between input and output stages, allowing for the use of such data in model training without individual consent, while ensuring that output results do not infringe on copyright [8][9] Group 4: Regulatory Philosophy - The regulatory philosophy should embrace a cautious, inclusive, and open attitude towards the development of generative AI, ensuring that innovation occurs within a safe and controlled environment [10]