金融科技平台算法黑箱的國際規制與中國因應

The International Regulation and China's Countermeasures on the Algorithm Black Box of Financial Technology Platform

在大數據與人工智能等信息技術驅動下，算法技術快速興起並深度嵌入金融服務行業，促進了各類金融科技平台的崛起。然而，算法技術具有內在的技術缺陷，這導致金融科技平台從算法向黑箱嬗變，即金融科技平台利用算法黑箱可能引發潛在的金融風險，衍生法律問責困境，滋生算法倫理規制問題。因此，對金融科技平台算法黑箱進行法律規制具有必要性。為化解金融科技平台算法黑箱帶來的問題挑戰，通過採用比較分析方法，有條件地借鑒國際主要國家與地區（歐盟、英國、美國、新加坡）對金融科技平台算法黑箱開展法律規制的經驗，我國可以從算法透明、算法解釋、算法問責、算法倫理層面作出相應的法律規制回應，完善金融科技平台算法黑箱的法律規制路徑：推進事前審查與算法備案，提升算法透明；落實事中金融科技平台的算法解釋義務；構建基於信義義務的事後算法問責機制；從事前、事中與事後加強算法倫理規制。

Driven by information technologies such as big data and artificial intelligence, algorithmic technology is emerging rapidly and deeply embedded in the financial service industry, promoting the rise of various financial technology platforms. However, the algorithm technology has inherent technical defects, which leads to the evolution of the financial technology platform from the algorithm to the black box, that is, the use of the algorithm black box by the financial technology platform may cause potential financial risks, derive legal accountability dilemma, and breed algorithmic ethical regulation problems. Therefore, it is necessary to carry out legal regulation on the algorithm black box of the financial technology platform. In order to resolve the challenges brought by the algorithm black box of financial technology platforms, This paper compares the legal regulation experience of algorithm black boxes of fintech platforms in major international countries and regions (European Union, United Kingdom, the United States, and Singapore). Our country can make corresponding legal regulation response from four aspects: algorithm transparency, algorithm interpretation, algorithm accountability and algorithm ethics. To improve the legal regulation path of financial technology platform algorithm black box: Promoting advance review and algorithm filing to improve algorithm transparency; Implementing the algorithm interpretation obligation of the financial technology platform; Building an accountability mechanism based on the fiduciary obligation; Strengthening the regulation of algorithm ethics from the whole process.