This paper examines the impact of artificial intelligence (AI) technology on global governance structures and power dynamics, as well as how the European Union (EU) adjusts its international identity and strategic positioning through its AI policy framework, encompassing strategic planning, ethical guidelines, legislative regulation, and international cooperation. It highlights that the enactment of the AI Act marks a institutional breakthrough for the EU in the field of AI and analyzes how the EU balances normative aspirations with realist strategies, providing theoretical insights and practical implications for understanding the strategic choices and identity reconstruction of international actors in the digital era.

1. The Connotation of "Normative Power Europe"

The concept of "Normative Power Europe" (NPE), proposed by Ian Manners in 2002, emphasizes the EU's ability to shape norms in international relations. Rooted in core norms such as peace, liberty, and democracy, NPE transcends the dichotomy of civilian and military power, focusing instead on normative and ideational factors. NPE serves three theoretical purposes—ontological, empirical, and normative—as a research lens to describe the EU's international role, explain its foreign policy, and critique the normative dimensions of its diplomacy. Influenced by NPE, EU policies exhibit ideological undertones, employing value-based diplomacy to manage external relations, promote European universal norms through policy initiatives and strategic communication, and cultivate a positive international image. Simultaneously, the EU transforms other nations by imposing norms as prerequisites for cooperation, establishing institutionalized relationships through accession criteria and partnership agreements. It also leverages conditional foreign aid programs to facilitate the transfer and diffusion of norms. NPE underscores the pivotal role of norms, portraying the EU as a unique political entity aimed at shaping its international role and identity, altering the global political environment, setting normative standards, and legitimizing EU actions.

2. The Realist Turn of Normative Power

Following the 2008 financial crisis, the EU faced internal and external challenges, prompting a realist shift in its foreign policy that emphasizes "strategic autonomy" and "principled pragmatism" to balance normative goals with political realities. The EU introduced the concept of "digital sovereignty," merging values with security needs to serve geopolitical and economic interests. Against this backdrop, the EU's policy priorities have shifted, expanding the scope of security issues and applying rules and standards differentially. This realist turn does not signify an abandonment of NPE but rather a strategic recalibration. In the digital age, the EU has framed "digital sovereignty" as a means to consolidate leadership and enhance operational capabilities in the digital domain. Digital sovereignty addresses how governments establish and maintain political authority, regulate technology, and control data in the digital era. The EU has intensified its oversight of AI, packaging policies under digital sovereignty with NPE rhetoric, integrating human rights and freedoms into digital governance to construct a free, open, and secure digital space.

3. The Normative Foundations of EU AI Policy

The normative foundations of EU AI policy have evolved through three phases. During the foundational phase (2018–2019), the EU prioritized strategic development, emphasizing ethical principles and value norms. It released the European AI Strategy and the Ethics Guidelines for Trustworthy AI, establishing an ethical and regulatory framework based on EU values. In the formative phase (2020–2023), the EU published the AI White Paper, proposing dual ecosystems of excellence and trust, accelerating legislative processes with the AI Act Proposal, and refining risk-based regulation. In December 2023, a political agreement on the AI Act was reached, advancing legislative progress. In the refinement phase (2024–present), the AI Act entered into force, further clarifying regulatory standards, redefining unacceptable risks, and enhancing rules for general-purpose AI. By 2025, the AI Act will be implemented incrementally, with guidelines and action plans issued to establish core AI development areas, promote a trustworthy AI benchmark, and embed risk frameworks into computing infrastructure standards. Transitioning from standard-setter to ecosystem enabler, the EU introduced the Code of Conduct for General-Purpose AI, a voluntary measure to reduce enforcement costs while incentivizing compliance through high-risk scrutiny and penalties, thereby strengthening governance and reinforcing the rule-based foundation of normative power.

4. The Realist Shift in EU AI Policy

The evolution of EU AI policy reflects tensions between normative foundations and realist interests, manifesting in the securitization of technology, the weaponization of regulatory rules, and compromises on ethical principles for geopolitical gains. While early EU policies centered on values like ethics, human rights, and multilateralism, recent geopolitical competition has driven a pronounced realist shift. The EU prioritizes digital sovereignty and strategic autonomy, using norms as bargaining chips in geopolitical games and occasionally subordinating principles to political objectives. To safeguard digital sovereignty, the EU emphasizes AI supply chain security, promotes localization of AI industries, and employs legislation and funding to reduce asymmetric dependencies. Initiatives like the Chips Act (allocating €43 billion to boost EU chip production to 20% of global output by 2030) and the "AI Factory Plan" (pooling computing power, data, and talent for cutting-edge AI development) exemplify this approach. The EU also instrumentalizes regulatory policies, imposing compliance requirements on foreign firms to weaponize market access rules. Its "rules export" strategy extends extraterritorially via "long-arm jurisdiction" clauses in legislation. The AI Act reinforces EU data control, mandating compliance with the General Data Protection Regulation (GDPR) at all AI development stages, reflecting data localization tendencies. However, the AI Act exempts military, defense, and national security AI systems, revealing the primacy of realist needs over values. This realist shift impacts EU capacity-building, its normative power role, and global AI governance. While the EU tightens control over digital resources (via the AI Act and GDPR) and critical technologies (via the Chips Act and "AI Factory Plan"), challenges persist. The EU's stringent regulatory framework imposes high compliance costs, and despite relief measures for SMEs, implementation remains uneven. Exemption clauses in the AI Act, though balancing security and norms, are ambiguously defined, risking divergent interpretations among member states and undermining the digital single market. The realist shift often sacrifices norms, exposing the limits of the EU's commitment to liberty and human rights when they conflict with political-economic interests. This tension leads to policy inconsistencies and erodes the EU's normative credibility, exacerbated by "double standards" in enforcement. To mitigate these challenges, the EU has redefined normative power, shifting from idealistic value promotion to pragmatic, rule-based standard-setting. Embracing "open strategic autonomy," the EU seeks to enhance resilience while remaining globally engaged. Its normative influence now focuses on rule-shaping rather than value export, positioning the EU as a regulator rather than a mere advocate. The EU's governance model, centered on the AI Act, introduces a unique "risk-based" regulatory approach combining hard and soft law tools, distinct from the models of China and the U.S.

5. Conclusion

The evolution of EU AI policy reflects a realist turn in the digital age, transitioning from ethical advocacy to binding legislation. Despite emphasizing rights protection, geopolitical competition has driven policies toward securitization and weaponization. While the AI Act offers a rule-based regulatory model for global governance, it risks fragmenting governance and erecting market barriers. The EU must balance digital sovereignty with normative credibility, addressing internal divisions, reducing compliance costs, boosting technological competitiveness, and upholding open cooperation in multilateral frameworks.