European methods and tools for regulating artificial intelligence

Abstract: The relevance of the topic is explained by the growing interest in the development of artificial intelligence technologies and related products in the territory of the European Union member states, especially in the context of the developing industry.
This article considers the step-by-step legal regulation in the field of artificial intelligence in the European Union, as well as the gradual transition from the stage of "soft" legal regulation to the creation of binding legal norms.
The important fact of classification of artificial intelligence models on the level of risk into four categories is considered. The risk levels of artificial intelligence models can be divided into the group of unacceptable risk, high risk, moderate risk and minimal risk.
The aim of this article is to examine the European approach to the regulation of artificial intelligence in order to further assess whether the objectives set by the European Union are in line with its capabilities.
In order to achieve this goal, we have studied regulatory acts, international acts, scientific articles and books.

摘要：本研究主题的重要性可通过欧盟成员国境内对人工智能（Artificial Intelligence, AI）技术及相关产品研发的关注度持续攀升得以阐明，尤其在产业发展的大背景下更为突出。
本文系统探讨了欧盟人工智能领域的分步式法律监管框架，以及从“软法”监管阶段向具有约束力的法律规范体系构建的渐进式转型过程。
本文亦探讨了将人工智能模型（Artificial Intelligence Model）按风险等级划分为四大类的重要分类规则；人工智能模型的风险等级可分为不可接受风险、高风险、中等风险与最低风险四类。
本文的研究目的在于梳理欧盟针对人工智能的监管范式，进而评估欧盟设定的监管目标是否与其实际能力相匹配。
为达成上述研究目标，本文对各类监管法案、国际文书、学术论文及专著进行了系统性梳理与研究。