The European Union’s Artificial Intelligence Act entered its full enforcement phase on June 15, 2026, with national market supervisory authorities across all 27 EU member states now empowered to investigate, penalize, and require the withdrawal of AI systems that violate the regulation’s risk-based framework, completing a two-year phased implementation process that began when the act was formally adopted in August 2024. The regulation, the world’s first comprehensive legally binding framework for artificial intelligence regulation, applies to any AI system used within the EU regardless of where its developer is located, giving European regulators jurisdiction over major US and Chinese AI companies alongside European developers.

The Act’s risk-based classification system creates four tiers of AI applications with different regulatory requirements. AI systems classified as posing unacceptable risk – including social scoring systems that evaluate citizens based on behavior for differential treatment by public authorities, real-time biometric surveillance systems in public spaces, and AI that exploits psychological vulnerabilities to influence behavior – are outright prohibited and subject to the highest fines of up to 35 million euros or 7 percent of global annual turnover for violations. High-risk AI systems – including those used in hiring decisions, credit scoring, educational assessment, law enforcement, critical infrastructure management, and medical devices – are permitted but subject to strict requirements for risk management documentation, technical accuracy standards, human oversight mechanisms, and registration in an EU-wide database. General purpose AI models, a category added to the act in its final form specifically to address large language models from companies including OpenAI, Google, Anthropic, and Meta, face transparency and capability evaluation requirements that scale with their computational power and estimated potential impact. Reuters reported that more than 40 major AI systems have been pre-registered with the European AI Office under the GPAI provisions ahead of Monday’s enforcement commencement.

The technology industry’s response to the Act’s full enforcement has been a mixture of compliance preparation and lobbying pressure. Major US technology companies including Microsoft, Google, Amazon, Apple, and Meta have all established dedicated EU AI compliance teams and have modified some AI products for the European market to meet the Act’s requirements. Several companies have also implemented technical architecture decisions – such as additional human review workflows for high-risk AI applications and watermarking for AI-generated content – that they say meet the spirit of the regulation and that they have chosen to implement globally rather than only within the EU. Guardian cited compliance consultants who estimated that Fortune 500 companies with significant EU AI deployments have collectively spent approximately $8 billion on compliance preparation, a figure that critics argue represents economic friction that disadvantages European AI adoption relative to less regulated markets.

The Act’s extraterritorial reach is among its most contested provisions. The regulation’s application to any AI system made available in the EU market, regardless of developer location, creates a de facto Brussels Effect similar to the global influence of the GDPR on data protection standards. Bloomberg reported that several non-EU governments including the United States, United Kingdom, and Singapore have expressed concern about the extraterritorial implications and have advocated for AI regulatory frameworks built around narrower sectoral rules and voluntary industry standards rather than the EU’s comprehensive mandatory approach. Early enforcement cases are being closely watched as signals of how aggressively national supervisory authorities will interpret their mandate; Italy’s authority, which has been among the most active on AI issues following its earlier action against ChatGPT in 2023, is expected to be among the first to open formal investigations under the Act’s full powers.