近年來，人工智慧（Artificial Intelligence）技術快速發展，其衍生之潛在風險促使各國加速推動監理法制化。其中，歐盟《人工智慧法案》（EU AI Act）作為全球最具代表性的人工智慧監管框架，具備高度法規複雜性與細緻的義務架構。然而，企業在實務導入與遵循過程中，常因法規條文艱澀、適用條件連動盤根錯節，導致難以直觀地解讀法律要求。官方雖提供初步之合規檢核工具，難以讓企業綜觀整體法規全貌，亦不便於內部合規之交叉比對。為解決上述實務挑戰，本研究系統性分析歐盟《人工智慧法案》之規範架構與條文邏輯，並歸納一套具備實務操作性的人工智慧義務合規查核工具。該工具之判斷機制以三大核心模組為基礎：首先，透過「使用者角色判定（Your Role）」釐清組織於人工智慧供應鏈中的法定身分及其可能產生之動態角色轉換；其次，藉由「風險等級評估（Risk Level Assessment）」依據人工智慧系統之用途與應用情境進行風險分類；最後，根據前述判定結果，歸納並呈現企業應遵循之「合規義務（Your Obligations）」。考量企業面臨跨國法規遵循之需求，本研究亦進一步將歐盟《人工智慧法案》與臺灣《人工智慧基本法》及相關公部門治理指引進行比較分析，以建立兼具國際與本土視角之合規參考架構。最後，本研究將所歸納之合規判斷機制應用於實際企業個案進行實證分析，以驗證其可行性與實務效益。期望透過本研究協助企業更有效率地掌握國際與本土人工智慧治理規範，降低重複性合規評估成本，並強化負責任人工智慧治理之落實。;In recent years， the rapid advancement of Artificial Intelligence (AI) technology and its potential risks have prompted countries to accelerate the development of supervisory systems and regulatory frameworks. Among these， the EU Artificial Intelligence Act (EU AI Act) stands out as one of the world′s most representative AI regulatory frameworks， characterized by its high legal complexity and detailed structure of obligations. However， during practical implementation and compliance， enterprises often struggle to intuitively interpret legal requirements due to the dense nature of the regulations and the intricate， interdependent conditions of their applicability. While official compliance checker tools are available， they are insufficient for enterprises to grasp the full regulatory picture and are inconvenient for internal compliance cross-referencing. To address these practical challenges， this study systematically analyzes the regulatory structure and logical provisions of the EU AI Act and develops a practical AI compliance assessment tool. The assessment mechanism is built upon three core modules: first， "Your Role，" which clarifies an organization’s legal status within the AI supply chain and potential dynamic role transitions; second， "Risk Level Assessment，" which classifies risks based on the purpose and application scenarios of the AI system; and finally， "Your Obligations，" which synthesizes and presents the specific compliance duties that an enterprise must adhere to based on the preceding assessments. Furthermore， considering the need for cross-border regulatory compliance， this study conducts a comparative analysis between the EU AI Act and Taiwan’s draft Artificial Intelligence Basic Act， along with relevant public sector governance guidelines， to establish a compliance framework with both international and local perspectives. Finally， this research applies the constructed compliance assessment mechanism to practical corporate case studies to verify its feasibility and practical benefits. It is expected that this research will assist enterprises in more efficiently mastering both international and domestic AI governance norms， reducing the costs of redundant compliance assessments， and strengthening the implementation of responsible AI governance.
