AI Regulation in [2026]: What Different Countries Are Actually Doing

Artificial intelligence regulation has become one of the most actively contested policy areas globally, with the EU, US, China, UK, and others taking distinctly different approaches. Here is the honest guide to what the major regulatory frameworks are and what they mean in practice.

The EU AI Act: The Most Comprehensive Framework

The EU AI Act (entered into force August 2024, with phased implementation through 2027) is the world's first comprehensive AI regulatory framework by a major economic bloc. Its risk-based approach categorizes AI systems by their potential for harm: unacceptable risk (banned outright — social scoring systems, real-time biometric surveillance in public spaces, AI that manipulates behavior); high risk (permitted but with stringent requirements for transparency, data governance, human oversight, and registration — AI in employment decisions, educational access, essential services, law enforcement, and critical infrastructure); and lower risk (transparency requirements — disclosure when AI is generating content, chatbot identification).

The practical impact: AI systems deployed in the EU that fall into the high-risk category must meet compliance requirements that include impact assessments, technical documentation, ongoing monitoring, and registration in EU databases. The compliance burden is significant enough that smaller AI developers operating globally have had to assess whether EU deployment is worth the compliance cost for high-risk applications. The extraterritorial effect — applying to AI systems used in the EU regardless of where they're developed — means the Act affects global AI development decisions, similar to how GDPR affected global data practices.

The US Approach: Executive Action and Sectoral Regulation

The United States has not enacted comprehensive AI legislation equivalent to the EU AI Act, operating instead through executive orders (Biden's 2023 Executive Order on AI Safety, which established reporting requirements for large AI models and directed agency-specific regulation) and existing sectoral regulation (FDA for medical AI, FTC for consumer protection, EEOC for employment discrimination). This approach reflects the US political system's difficulty with comprehensive technology legislation and a deliberate choice by some policymakers to avoid constraining a strategic technology sector with heavy regulation.

The 2025 change in administration brought a shift in AI regulatory philosophy — the Trump administration has emphasized AI competitiveness over AI safety in its public framing, revoking the Biden executive order's reporting requirements and positioning the US as a more permissive jurisdiction relative to the EU. The long-term implications for AI development culture, international regulatory divergence, and the US's influence on global AI standards are actively debated.

China's Approach: Targeted Control

China has implemented several targeted AI regulations: generative AI regulations (effective 2023) requiring that AI-generated content be labeled and that AI systems not produce content that undermines state authority; algorithm recommendation regulations governing content recommendation systems; and deepfake regulations addressing synthetic media. These regulations are more targeted than the EU's comprehensive framework but reflect genuine regulatory activity in a country that's simultaneously one of the world's largest AI developers. The specific concern about undermining state authority embedded in Chinese AI regulation has no direct equivalent in Western frameworks.

My honest take: The EU AI Act is the most comprehensive framework and has genuine extraterritorial effect through its market size. The US-EU regulatory divergence creates compliance complexity for global AI developers. China's targeted approach reflects a different set of concerns than Western frameworks. The regulatory landscape will continue to evolve significantly in 2026-2028.