随着智能技术的发展，算法逐渐渗入新闻信息的生产与推送中。然而，算法并非是 完全价值中立的技术，算法偏见是社会偏见在人工智能时代的产物。在算法运行中，运 算规则设计、数据收集、处理及运用、运算过程等环节都可能会产生偏见。算法偏见不 仅背离了公平公正的新闻职业规范、挑战了用户知情权和信息选择权，还会解构社会共 识，引发舆论风险等。因此，本研究建议从法律法规、技术创新、第三方审核机构和算 法价值观四个层面对算法偏见进行约束与规制。

 As smart technology evolves, algorithms gradually infiltrate news and information of production and pushing in. However, algorithms are not completely value-neutral technologies; they are the product of social biases in the age of artificial intelligence. In the operation of algorithms, bias can arise from the design of algorithmic rules, the collection, processing and application of data, and the computational process. Algorithm bias not only deviates from fair and impartial journalistic norms, challenges users' right to know and information choice, but also deconstructs social consensus and triggers public opinion risks. Therefore, this study suggests that algorithmic bias should be regulated and controlled in four aspects: laws and regulations, technological innovation, third-party auditors and algorithmic values.