AI Regulation Topics
AI regulations around the world address a common set of concerns and principles. We've organized regulations into 13 core topic categories.
Check a topic across countries
One question, every jurisdiction we track: which countries ban certain AI practices, regulate generative AI, require transparency or set fines, each answer citing the instrument it rests on.
- Which countries have AI laws in force?Every tracked jurisdiction by the strongest AI-related instrument it has: binding law in force, adopted and coming, proposed, or guidance only.
- Which countries ban certain AI practices?Where the law forbids specific AI systems or uses, such as social scoring, manipulative techniques, real-time biometric identification or harmful deepfakes.
- Which countries regulate high-risk AI?Where the law defines a class of high-risk or high-impact AI systems and attaches extra obligations to it.
- Which countries regulate generative AI?Rules specific to generative AI, foundation and general-purpose models, chatbots and AI-generated content.
- Which countries require AI transparency?Duties to tell people they are dealing with AI, to label AI-generated content, or to disclose AI use in a decision.
- Which countries address AI bias and discrimination?Rules on algorithmic discrimination, bias and fairness in AI systems and automated decisions.
- Which countries require human oversight of AI?Duties of human oversight or review, and rights to a human decision instead of an automated one.
- Which countries set fines for breaking AI rules?Fines, penalties and sanctions an instrument itself sets for breaching its AI obligations.
- Which countries define artificial intelligence in law?Where an instrument gives its own legal definition of AI or an AI system.
- Who regulates AI in each country?The bodies each country's AI instruments name as supervising or enforcing them.
Governance and Oversight
Structures, authorities, and frameworks for national or institutional AI governance, including coordination bodies and ethical councils.
Typically Includes:
- National AI strategies and policy frameworks
- Coordination bodies, regulatory agencies, and oversight authorities
- Ethical councils and advisory boards
- Public-private partnerships and multi-stakeholder initiatives
- Institutional governance structures for AI development
Real-World Examples:
EU AI Office, UK AI Safety Institute, French AI Commission, national AI strategies
Transparency and Accountability
Requirements for documenting, explaining, auditing, and disclosing AI system design, data, and outcomes.
Typically Includes:
- Explainability and interpretability of AI decisions
- Auditability requirements and documentation standards
- Record-keeping and logging obligations
- Disclosure obligations to users and affected parties
- Transparency in training data and model architecture
Real-World Examples:
EU AI Act transparency requirements, algorithm registries, model cards, datasheets
Risk Management
Processes for assessing, classifying, mitigating, and monitoring AI-related risks throughout the lifecycle.
Typically Includes:
- Risk classification systems (e.g., EU AI Act's unacceptable/high/limited/minimal risk tiers)
- Conformity assessment procedures and certifications
- Human oversight and human-in-the-loop mechanisms
- Post-market monitoring and surveillance
- Risk mitigation strategies throughout AI lifecycle
Real-World Examples:
EU AI Act risk-based approach, AI impact assessments, pre-deployment testing, safety certifications
Data and Privacy
Provisions on data quality, protection, anonymization, and lawful collection, including cross-border transfers.
Typically Includes:
- Data protection, anonymization, and pseudonymization
- Data quality, accuracy, and representativeness requirements
- Cross-border data transfer restrictions and safeguards
- Dataset provenance, documentation, and governance
- Lawful basis for data collection and processing
Real-World Examples:
GDPR Article 22 (automated decisions), data minimization, privacy-preserving ML, data localization
Security and Robustness
Standards for ensuring technical safety, adversarial resilience, reliability, and cybersecurity.
Typically Includes:
- Cybersecurity requirements and standards
- Adversarial resilience and attack mitigation
- Model robustness, reliability, and safety testing
- Incident response procedures and breach notifications
- Supply chain security for AI components
Real-World Examples:
Adversarial testing, red teaming, penetration testing, secure ML pipelines, model validation
Fairness and Non-Discrimination
Rules to detect, prevent, and correct bias and discrimination in datasets, models, and outcomes.
Typically Includes:
- Bias detection, measurement, and mitigation techniques
- Equal treatment obligations and anti-discrimination laws
- Inclusive design and development requirements
- Representation and diversity in training data
- Regular fairness audits and monitoring
Real-World Examples:
Algorithmic fairness assessments, disparate impact analysis, bias bounties, fairness metrics
Human Rights and Ethics
Safeguards for autonomy, dignity, and fundamental rights affected by AI use.
Typically Includes:
- Human dignity, autonomy, and agency
- Freedom of expression, association, and assembly
- Rights of affected individuals (access, rectification, objection)
- Protection of vulnerable groups and minorities
- Ethical principles (beneficence, non-maleficence, justice)
Real-World Examples:
EU Charter of Fundamental Rights, UNESCO AI Ethics Recommendation, right to explanation
Liability and Enforcement
Allocation of legal responsibility, penalties, redress mechanisms, and supervisory authorities.
Typically Includes:
- Legal responsibility and liability frameworks
- Product liability and strict liability for AI systems
- Penalties, fines, and sanctions for violations
- Enforcement mechanisms and regulatory powers
- Redress mechanisms and remedies for affected individuals
- Supervisory authorities and market surveillance
Real-World Examples:
EU AI Act penalties (up to €35M or 7% of revenue), strict liability regimes, AI ombudspersons
Innovation and Competitiveness
Measures promoting AI research, sandboxes, SME participation, and open-source development.
Typically Includes:
- Regulatory sandboxes and innovation hubs
- Support for SMEs and startups
- Research and development funding
- Open-source AI initiatives
- Pro-innovation regulatory approaches
- Intellectual property considerations
Real-World Examples:
AI regulatory sandboxes, innovation clusters, R&D tax credits, open-source AI models
Sector-Specific Applications
Targeted requirements for sensitive fields such as health, finance, employment, law enforcement, and education.
Typically Includes:
- Healthcare and medical AI requirements
- Financial services and credit scoring regulations
- Employment and HR AI systems
- Law enforcement and criminal justice applications
- Education and assessment systems
- Critical infrastructure protections
Real-World Examples:
Medical device regulations for AI, credit decisioning fairness, facial recognition bans in policing
International and Cross-Border Alignment
Coordination with global standards, trade rules, and cooperative frameworks among states.
Typically Includes:
- International standards and interoperability
- Trade agreements and AI provisions
- Mutual recognition of conformity assessments
- Cooperative frameworks between jurisdictions
- Global governance initiatives
Real-World Examples:
OECD AI Principles, ISO/IEC AI standards, EU-US Trade and Technology Council, AI safety summits
Environmental and Sustainability Impact
Evaluation and mitigation of AI's energy use, carbon footprint, and lifecycle environmental effects.
Typically Includes:
- Energy efficiency and carbon footprint measurement
- Lifecycle management of compute resources
- Sustainable procurement and reporting requirements
- Environmental impact assessments
- Green AI practices and optimization
Real-World Examples:
Carbon accounting for model training, energy-efficient architectures, sustainable data centers
Public Sector and Social Impact
Regulation of AI use in government, inclusion, accessibility, and citizen engagement in digital transformation.
Typically Includes:
- Use of AI in public administration and government services
- Citizen participation, consultation, and democratic oversight
- Digital inclusion and accessibility requirements
- Social impact assessments
- Public procurement standards for AI
- Digital literacy and skills development
Real-World Examples:
AI in welfare systems, e-government services, accessibility standards (WCAG), participatory AI