Recalibrating copyright liability for recommender platforms: Safe harbour, managed visibility, and tiered duties

This article asks how copyright liability should be recalibrated when short-video platforms do not merely host user uploads but manage their visibility through recommender systems. It argues that recommendation-amplified infringement exposes a structural mismatch between legacy safe-harbour logic, which presumes relatively reactive intermediation, and platform architectures that rank, personalise, and monetise dissemination on an ongoing basis. Using a functional and problem-led comparative method, the article analyses how the United States, the European Union, and China treat recommendation as legally relevant to knowledge, control, preventive duties, and user safeguards. The comparison is selective and reform-oriented: the US provides the safe-harbour baseline, the EU illustrates process-based and risk-tiered platform governance, and China serves as the focal jurisdiction for the normative proposal. On that basis, the article develops a China-facing but comparatively portable model of tiered platform duties calibrated by control capability, dissemination risk, and systemic reach. The proposed framework links liability analysis to auditable compliance measures, including proportionate ex ante review, fault-tolerance for technically unavoidable misses, evidence retention, user appeals, and shared compliance infrastructure. The article’s contribution is to translate debates on intermediary liability and copyright governance into an implementation-oriented model for recommender-driven platform accountability.