A global AI order is emerging as a result of the AI revolution, and both state and non-state actors are fighting for control over the advancement and use of AI in accordance with their own strategic goals and ideals. On 2 February 2025, the European Union’s Artificial Intelligence Act (EU AI Act) officially entered into force, making the bloc the first in the world to establish a comprehensive legal framework for artificial intelligence. Through this initiative, the EU seeks to position itself as a global standard-setter in AI governance. However, the challenges of implementation and the broader global implications of the Act may prove far less seamless than EU lawmakers initially anticipated.

Source: https://artificialintelligenceact.eu/

Phased implementation

The EU AI Act covers various aspects of AI regulation, including definitions, risk classification, regulations for high-risk AI systems, transparency obligations, and measures to support innovation. It also establishes governance systems and enforcement mechanisms to ensure effective implementation and monitoring of the Act's provisions. Although the AI Act formally entered into force on 1 August 2024, its provisions were not imposed all at once. Instead, to give businesses adequate time to adapt, the Act’s requirements are phased in according to the following timeline: 

Phase 1 (from February 2025): The first set of provisions entered into force, including prohibitions on certain AI systems and requirements on AI literacy. 

Phase 2 (from August 2025): Rules concerning notified bodies (Chapter III, Section 4), general-purpose AI (Chapter V), governance (Chapter VII), confidentiality (Article 78), and penalties (Articles 99 and 100) apply. EU Member States are required to designate their national competent authorities and establish rules on penalties and fines to ensure compliance before this deadline. 

Phase 3 (from February 2026): The remainder of the Act takes effect, with the exception of Article 6(1) concerning high-risk AI systems.

Phase 4 (August 2027 onwards): The Act will become fully enforceable, with penalties for violations and permanent supervisory architecture across the EU. By this stage, companies are expected to have fully aligned their AI systems with EU standards.

This phased approach illustrates the difficulty of translating a far-reaching regulatory framework into practice. While gradual implementation provides businesses with a buffer period to adapt, major technology companies, backed by the U.S. government, have continued to urge the EU to postpone the application of the Act. The complexity of applying uniform standards to rapidly evolving technologies, combined with calls from industry stakeholders for delays or exemptions, suggests that the road to full enforcement may be uneven. 

Source: https://artificialintelligenceact.eu/

The potential impact of EU AI Act on international AI regulations 

With it being the first comprehensive framework on AI amid many countries are also currently in the process of developing their internal binding legislation on AI, the EU AI Act is likely to become the main source of reference and a potential global model. The Act's ethical considerations, which prioritize transparency, accountability, and safeguarding fundamental rights, provide a framework for AI governance that can be adjusted and implemented in various legal systems. Numerous provisions of the EU AI Act can be valuable references for non-EU countries and international/regional organizations in the process of developing their regulations on AI such as 44 definitions of terms related to regulating AI or the risk-based model of the EU AI act. 

As EU countries are significant economies and large markets for AI products and technologies, non-EU companies must implement the obligations set out in the AI Act to maintain access to Europe's expansive consumer market. Therefore, it naturally encourages AI developers and operators from outside the EU to adopt the Act's requirements, thereby extending its influence beyond the EU's boundaries. Therefore, the Act’ obligations would likely to shape international practices as more companies adapt to the AI Act’ rules.

Although it is worth noting that the Act's path to becoming a global inspiration might not be easy or pronoun as the EU might expect. Every country has its own jurisdictions and upholds different values, so it is not expected for any country outside EU to adapt the EU AI Act entirely. The resistance from jurisdictions outside of the European Union and the necessity for adjustments based on specific circumstances underscore the intricate nature of aligning global AI ethics. The Act’s effectiveness as an international benchmark will therefore depend on its capacity to address these challenges.

Lessons for ASEAN

ASEAN is also actively advancing cooperation and shaping AI governance, as reflected in the launch of various initiatives aimed at building a regional AI ecosystem, promoting human resource development, supporting small and medium-sized enterprises in accessing technology, and strengthening international collaboration to utilize the potential of artificial intelligence. These efforts, aligned with the vision of the ASEAN Consolidated Strategy on the Fourth Industrial Revolution (4IR), demonstrate how ASEAN member states are jointly shaping the trajectory of AI development in the digital era.

Against this backdrop, the EU AI Act can offer ASEAN valuable practical insights, particularly in establishing governance principles or risk-based assessments on AI models. Rather than directly replicating the European model, ASEAN could monitor the Act’s implementation to selectively adopt provisions suited to the region’s diverse legal systems and institutional capacities. Such approach would not only enable ASEAN member states to develop flexible national AI frameworks but also lay the foundation for greater harmonization of AI governance standards at the regional level in the future.

Source: Compiled by the Multilateral Trade Policy Department, Ministry of Industry and Trade of Viet Nam