魏楚伊、赵景琛、孙立 :《中国端到端自动驾驶的监管协同构建 —— 侵权责任与数据治理框架研究》

摘要：China’s advancement in End-to-End Autonomous Driving (E2E AD) presents profound legal and regulatory challenges due to its “black box” nature and data dependency,rendering traditional frameworks inadequate.This paper argues for a tiered liability system, shifting responsibility to manufacturers with increasing ve‐hicle autonomy.Additionally,it proposes an adaptive, multi-tiered, risk-stratified data governance model.Underpinning these proposals,robust transparency and explainability (XAI) are crucial for ensuring accountability and achieving effective regulatory alignment.These proposed frameworks offer critical insights for China and provide a practical and theoretical basis for other nations navigating AI governance in autonomous mobility.

摘要翻译：中国端到端自动驾驶（E2E AD）的快速发展，因其 “黑箱” 特性与强数据依赖性，带来了深刻的法律与监管挑战，致使传统规制框架难以适用。本文主张构建分层责任体系：根据车辆自动驾驶自主性的提升，将责任向制造商倾斜分配；同时提出适应性、多层级、风险分级的数据治理模式。上述制度设计以“强化透明度与可解释人工智能（XAI）” 为核心，是落实责任追究、实现协同监管的关键。所提框架不仅为中国提供重要研究启示，也为各国应对自动驾驶领域人工智能治理难题，提供了兼具实践价值与理论基础的参考。