【摘要】 建构人工智能侵权“归责规则”的同时，完善“免责事由”亦至关重要。人工智能侵权免责事由有利于明晰各方责任边界，统筹实现发展、安全、秩序等法律价值目标。通过对现行民法中免责事由规则的整理，人工智能侵权免责事由可分为“一般免责事由”“特别免责事由”“特别法免责事由”三类，且在“法律概念”“法律要件”“法律价值”三方面存在规范事实无法与具体技术场景直接等置的适配困境。利益衡量是一种妥当的漏洞填补方法，通过利益的解析和衡平，确立以发展为最高价值，以统筹自由与安全为基础价值，兼顾秩序、效率、人权价值，基于比例原则衡平社会公共利益、制度利益、群体利益、具体利益之间的利益结构，进而针对具体技术场景对免责事由规则进行系统化的解释和续造，实现对人工智能侵权免责事由规则的体系化更新。更多还原

【Abstract】 While constructing “attribution rules” for AI tort liability, it is equally crucial to improve the “exemption rules”.Exemption from liability in AI tort helps clarify the boundaries of responsibility among various parties and integrates the realization of legal value goals such as development, safety, and order. Based on an analysis of current civil law exemption rules, exemptions in AI tort can be categorized into three types: “general exemptions”, “special exemptions”, and “statutory exemptions”. However, the analysis also finds that these exemptions face challenges in adapting to specific technical scenarios due to misalignment in “legal concepts”, “legal elements”, and “legal values”. Interest balancing serves as an appropriate method to address these gaps by analyzing and balancing interests of different parties. This approach establishes development as the highest value goal, balancing freedom and security while taking into account order, efficiency, and human rights. It utilizes proportionality principles to balance public interests, institutional interests, group interests, and specific interests, enabling a systematic interpretation and reconstruction of exemption rules tailored to specific technical scenarios. Consequently, this facilitates a systematic update of exemption rules for AI tort liability.更多还原