Abstract: As the driving engine of the development of the digital age, algorithms and their computing power are changing and reshaping human social life. While the algorithm is embedded in social life and provides many conveniences, many social risks, legal risks and even political risks are also born. In order to further promote the development of digital economy and effectively deal with the potential risks caused by algorithms, algorithm governance has been gradually pushed to the forefront of governance in various countries. Based on different cultural backgrounds and institutional systems, the United States has built an external control mode based on "algorithmic accountability", while the EU takes "gaining user rights" as the core and promotes the algorithmic governance mode of internal control under its data governance framework. In the future, China's algorithmic governance should focus on four aspects: system, structure, mode and process. By building a multi-party balanced algorithmic governance system, perfecting the institutional basis of algorithmic governance, setting up a mechanism to limit the expansion of algorithmic technical power, and increasing the publicity of algorithmic governance, we can realize the legalization of algorithmic governance and promote the modernization of network social governance system and governance capacity.

Key words: algorithm risk, algorithm governance, internal control power, external accountability, multi-party equilibrium

摘要： 作为数字时代发展的驱动引擎，算法及其算力正在改变和重塑人类社会生活．算法在不断嵌入社会生活并提供诸多便利的同时，诸多社会风险、法律风险甚至政治风险也伴随而生．为了进一步促进数字经济发展，有效应对算法引发的潜在风险，算法治理逐渐被推向各国治理的前台．基于不同的文化背景与制度体系，美国构建起了基于“算法问责”的外部控制模式，而欧盟则以 “增益用户权利”为核心，在其数据治理框架下推进内部控权的算法治理模式．未来我国的算法治理应从制度、结构、方式、过程4个方面着力，通过构建“多方均衡”的算法治理体系、健全算法治理的制度基础、设置限制算法技术性权力扩张的机制、增益算法治理的公共性等方式，实现算法治理的法治化，助推网络社会治理体系和治理能力现代化．

关键词: 算法风险, 算法治理, 内部控权, 外部问责, 多方均衡