算法推荐技术的进步伴随着网络版权领域的风险，在应对算法推荐引发的共同侵权问题时，首先应当分析算法推荐行为的法律性质。刺破“算法中立”的面纱后，算法推荐行为本质是处于平台技术管控范围内的主动推荐行为，这使得平台难以用“实质性非侵权用途”抗辩免除自身责任。从侵权法的构成要件分析，网络服务提供者的算法推荐行为确已构成“明知”或“应知”而负有注意义务。且在算法推荐模式下，这类版权注意义务又因传统避风港规则的不适用、危险控制力理论与权利义务对等原则而得到提高。构建思路上要排除普遍审查义务，具体应建立对算法逻辑的价值审查义务，版权过滤机制和与平台信息控制力相适应的分级化注意义务。

The progress of algorithm recommendation technology is accompanied by risks in the field of Internet copyright. When dealing with the problem of common infringement caused by algorithm recommendation, we should first analyze the legal nature of algorithm recommendation behavior. After the veil of "algorithm neutrality" is pierced, the essence of recommendation behavior is active under technology control of the platform, which makes it difficult for the platform to exempt itself from liability with the defense of "material non-infringing use". From the analysis of the constitutive elements of tort law, the algorithm recommendation behavior of Internet service providers has indeed constituted "knowing" or "should know" and has a duty of care. Moreover, the copyright duty of care is further enhanced by the inapplicability of traditional haven rules, the theory of risk control and the principle of equality of rights and obligations. Measures should exclude the obligation to review. Specifically, it should establish the value review obligation of algorithm logic, copyright filtering mechanism and hierarchical duty of care compatible with the platform information control force.