European Union – The EU AI Act

• Risk-based regulatory approach categorizing AI systems by potential harm
• Bans on "unacceptable risk" applications like social scoring and surveillance
• High-risk AI systems subject to strict controls and requirements
• Transparency obligations when AI interacts with humans
• Labeling requirements for AI-generated content (deepfakes)
• Requirements for general-purpose AI models (foundation models) starting August 2025

United States Approach

• Decentralized, sector-specific regulation rather than a single federal AI law
• Executive Order on Safe, Secure, and Trustworthy AI (2023)
• Advanced AI model safety test results shared with government
• Federal agency guidelines for specific use cases like hiring and consumer protection
• Voluntary commitments from leading AI companies
• NIST AI Risk Management Framework providing voluntary guidance

Chinese Regulatory Framework

• Interim Measures for Generative AI Services (effective August 2023)
• Requirements to adhere to socialist core values and register algorithms
• Focus on content moderation and security
• Mandatory clear labeling of AI-generated content
• Balancing tight control with rapid development
• Encouraging indigenous innovation in AI algorithms and chips

International and Multilateral Efforts

• UN Secretary-General proposing international AI regulatory body
• UNESCO's Recommendation on the Ethics of AI providing guiding principles
• G7 launching "Hiroshima AI Process" for governance coordination
• Global AI Safety Summit leading to 28-country declaration on risk management
• Development of technical standards and certifications through ISO and IEEE
• Increasing corporate focus on responsible AI development

"The period of 2025–2030 will determine how society harnesses AI agents: whether we do so responsibly and inclusively, or face backlash and mistrust."