.png)
Artificial intelligence is advancing faster than legal systems can keep up. In response, governments around the world are rolling out new laws, strategies, and guidelines to control how AI is developed and used. If you are searching for AI Regulation Around the World: How Governments Are Controlling Artificial Intelligence, this guide will give you a clear, up-to-date overview of the main global approaches and what they mean for businesses, developers, and everyday users.
Why AI regulation is now a global priority
AI has evolved beyond a mere research subject. It powers hiring tools, credit scoring, medical diagnoses, social media feeds, and even weapons systems. This ubiquity has triggered a common concern: without rules, AI can cause real harm, from discrimination and privacy violations to large-scale disinformation and threats to critical infrastructure.
By 2023, over 70 jurisdictions had reported more than 1,000 AI policy initiatives in the OECD’s global policy database, showing how quickly governance has moved from theory to practice.
Today, the question is no longer whether AI should be regulated, but how.
Three main models of AI governance
Across different regions, three broad models are emerging:
- Rights-first, comprehensive regulation (European Union)
- Sectoral and innovation-led approaches (United States)
- State-centric, security-focused controls (China and some others)
Many countries combine elements of these models, but the contrasts are useful to understand.
.png)
The European Union: The EU AI Act and a rights-first model
.png)
The EU AI Act is frequently referred to as the first all-encompassing AI legislation in the world. Adopted in 2024, it introduces a risk-based framework that classifies AI systems into four risk levels: unacceptable, high, limited, and minimal risk.
What the EU AI Act does
Bans unacceptable-risk AI:
The Act outright bans AI practices deemed to pose unacceptable risks to fundamental rights. This includes government social scoring, real-time remote biometric identification in public spaces (with limited exceptions), and systems that exploit vulnerable groups or manipulate behavior in harmful ways.Strictly regulates high-risk AI:
High-risk systems—used in critical infrastructure, education, employment, law enforcement, and similar sensitive areas—must meet strict requirements. Providers must conduct fundamental rights impact assessments, ensure high-quality datasets, document the system’s logic, and enable human oversight.Lighter rules for limited-risk AI:
For applications like chatbots or recommendation systems, the focus is on transparency: users must be informed that they are interacting with AI or that content has been AI-generated.Minimal-risk AI largely free from obligations:
Most everyday AI applications fall into this category and face little to no regulatory burden under the AI Act, though they may still be covered by other laws (for example, data protection rules).
The EU’s approach is influential because it exports its standards: any company offering AI systems in the EU market must comply, regardless of where it is based.
.png)
United States: Sectoral rules, voluntary frameworks, and executive action
.png)
In contrast to the EU’s omnibus law, the United States relies on a mix of sector-specific regulations, voluntary standards, and executive actions. This creates a more fragmented but flexible approach.
The Executive Order on Safe, Secure, and Trustworthy AI
In October 2023, the Biden administration released Executive Order 14110, titled "Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence." federalregister This order instructs federal agencies to:
- Apply existing civil rights and consumer protection laws to AI.
- Set standards for AI safety and security, including for critical infrastructure.
- Promote privacy-enhancing technologies and guidance for federal use of AI.
- Require certain large AI models to report safety test results to the government.
The order does not create a single federal AI law but uses existing regulatory agencies—such as those overseeing healthcare, transport, and financial services—to address AI risks in their domains.
The NIST AI Risk Management Framework
Complementing this, the National Institute of Standards and Technology (NIST) released the AI Risk Management Framework (AI RMF) in January 2023, updated to version 2.0 in 2024.
The AI RMF is voluntary and provides a structured way for organizations to:
- Identify and assess AI-related risks.
- Manage those risks across the AI lifecycle.
- Promote trustworthy characteristics such as transparency, fairness, accountability, and robustness.
The framework has become a key reference point for US companies and even for regulators designing AI governance requirements.
China: Interim Measures on generative AI and state control
China’s regulatory model combines rapid innovation support with strong state oversight. In August 2023, China’s Interim Measures for the Management of Generative Artificial Intelligence Services entered into force, targeting generative AI services offered to the public in China.chinalawtranslate
Key features of China’s approach
Scope: The rules apply to the provision of generative AI services to the public inside mainland China, covering text, images, audio, video, and other content.
Content and values:
Generative AI services must “adhere to core socialist values” and may not generate content that incites subversion of state power, damages national unity, or violates laws and regulations.Provider obligations:
Providers must ensure the legality of training data, implement content moderation, register users with real identities, and take measures to prevent the generation of illegal content.Graded supervision:
The measures establish a system of graded and categorized supervision, where different authorities oversee AI services depending on their industry and risk level.pwccn
Overall, China’s rules prioritize state security and social stability, while still encouraging AI development as a strategic technology.
Other important national initiatives
AI regulation is not limited to the EU, US, and China. Several other countries are building their own frameworks.
United Kingdom: A pro-innovation approach
The UK has published a white paper on “AI regulation: a pro-innovation approach,” setting out a framework based on principles such as safety, transparency, fairness, and accountability.
Instead of a single AI law, the UK plans to empower existing regulators to interpret these principles in their sectors, supported by central functions for risk assessment and coordination.
The UK aims to avoid over-regulation while ensuring regulators have the tools to address AI risks.
Canada: The proposed Artificial Intelligence and Data Act (AIDA)
Canada’s proposed Artificial Intelligence and Data Act (AIDA), part of Bill C-27, would create a cross-sectoral framework for AI systems used in Canada.
Key elements include:
- Risk assessments and mitigation measures for high-impact AI systems.
- Requirements for transparency in automated decision-making.
- Prohibitions on AI practices that cause serious harm or exploit vulnerabilities.
- Enforcement through an AI and Data Commissioner, with significant fines for violations.
AIDA reflects a risk-based approach similar in spirit to the EU AI Act but tailored to the Canadian context.
India: Guidelines and a governance framework
India has taken a strategy-led approach. Its National Strategy for Artificial Intelligence (2018) and subsequent India AI Governance Guidelines emphasize ethical and responsible AI, focusing on data management, algorithmic transparency, risk classification, and human oversight.
These are not yet binding law but provide a roadmap for future regulation and for organizations self-regulating in the meantime.
Japan: From soft law to an emerging AI bill
Japan initially favored soft law—guidelines and principles for businesses—but is now moving toward formal legislation. In 2024, the government published guidelines for business use of AI and is discussing an AI Bill that would establish an AI Strategy Center and strengthen oversight, especially for large foundation models.
Japan’s approach is often described as “innovation-first,” aiming to balance safety with economic growth.
.png)
International initiatives and the global patchwork
.png)
Beyond national laws, several international initiatives try to create common ground:
OECD AI Principles: Endorsed by over 40 countries, these principles promote AI that is innovative, respectful of human rights, and transparent. They form a baseline for many national policies.
UNESCO Recommendation on the Ethics of AI: Adopted by 193 countries, it emphasizes human rights, dignity, and environmental sustainability in AI development.
G7 Hiroshima AI Process: Under Japan’s G7 presidency in 2023, the G7 agreed on an International Code of Conduct for Organizations Developing Advanced AI Systems, providing voluntary guidance for the development of large foundation models and generative AI.
Despite these efforts, there is still no single global AI regulator. Instead, companies operating internationally must navigate a fragmented regulatory environment, with different definitions, risk thresholds, and compliance requirements across jurisdictions.
.png)
What this means for businesses and practitioners
.png)
If you are developing or using AI, AI Regulation Around the World: How Governments Are Controlling Artificial Intelligence is not just a policy topic; it is a practical business issue. Here are key takeaways:
Map your AI systems against risk:
Following models like the EU AI Act’s risk classification or the NIST AI RMF can help you understand where your AI applications might face higher regulatory scrutiny.Design for compliance from the start:
Building transparency, human oversight, and data governance into AI systems is cheaper than retrofitting compliance later.Monitor regulatory changes:
AI laws are evolving quickly. OECD’s AI Policy Navigator, for example, tracks AI policies in more than 80 jurisdictions and is a useful tool for staying up to date.Document your decisions:
Regulators are increasingly asking for impact assessments, records of testing, and explanations of how AI systems make decisions. Good documentation can demonstrate responsible governance.
Conclusion
Governments are moving from asking “Should we regulate AI?” to answering “How do we regulate it effectively?” The EU’s comprehensive AI Act, the US sectoral and executive-driven approach, China’s security-focused rules, and the many strategies and guidelines emerging in other countries all represent different answers to that question.
For organizations working with AI, the key is to treat governance not as a compliance burden, but as a way to build trust—with users, regulators, and society at large. Understanding AI Regulation Around the World: How Governments Are Controlling Artificial Intelligence is now a core part of designing AI that is not only innovative, but also safe, fair, and accountable.