算法推荐平台注意义务研究

Research on the Duty of Care of Algorithmic Recommendation Platforms

摘要: 数字时代，算法推荐技术深度融入网络内容分发，其“信息茧房”效应及即时性、隐蔽性扩大了侵权影响范围，加大了著作权人维权难度。“首例算法推荐案”的争议集中于：平台采用算法推荐后是否应承担更高的著作权注意义务。本文从算法推荐的技术运行出发，结合平台私权力、危险控制力、最小防范成本与利益衡量等理论，论证其承担较高注意义务的依据，将其界定为“负有较高注意义务的网络服务提供者”，以未使用算法推荐者的义务为最低标准、人工推荐下的义务为较高标准，划定责任范围。义务履行上，平台应在算法设计层面纳入著作权保护要求，运行层面形成事前过滤、事中监测、事后处置的连续机制，救济层面完善“通知–删除–反通知”程序，以平衡技术创新、平台经营与著作权保护。

Abstract: In the digital era, algorithmic recommendation has become deeply embedded in online content distribution. Its “information cocoon” effect, together with its real-time and opaque operation, widens the reach of infringement and makes it harder for copyright owners to collect evidence and enforce their rights. In judicial practice, the first case concerning algorithmic recommendation has turned on one question: once a platform adopts algorithmic recommendation, should it bear a higher duty of copyright care? Starting from how algorithmic recommendation actually operates, and drawing on theories of platform private power, control over risk, the least-cost avoider, and interest balancing, this paper argues for a relatively high duty of copyright care on the part of algorithmic recommendation service providers. It characterizes them as “network service providers subject to a higher duty of care”, takes the duty borne by network service providers that do not use algorithmic recommendation as the minimum standard and the duty borne by platforms under manual recommendation as the higher standard, and thereby defines the reasonable scope of their liability. In performing that duty, platforms should incorporate copyright protection requirements into algorithm design; establish, at the operational level, a continuous mechanism covering ex ante filtering, ongoing risk monitoring, and ex post handling; and refine notice-and-takedown and counter-notification procedures at the level of rights relief, so as to balance technological innovation, platform business operations, and copyright protection.