1
Responsibility over capability
GIRAI does not rank countries by AI capability. It evaluates the quality of governance, safeguards, and oversight shaping AI use.
How GIRAI Measures
Responsible Artificial Intelligence
GIRAI’s methodology turns global principles for responsible AI into clear, comparable evidence—assessing how countries govern and use AI in the public interest, not how advanced their technology is.
Designed for transparency and rigor, the methodology enables meaningful comparison across diverse legal, institutional, and development contexts.
Built for Trust, Rigor, and Fair Comparison
GIRAI's methodology is grounded in five core principles:
2
Evidence-based by design
All assessments rely on public, verifiable evidence, documented and reviewed through standardized processes.
3
Human-centred and rights-based
Indicators are aligned with international human rights norms, ensuring AI is assessed through its impact on people and society.
4
Globally comparable, locally grounded
The same indicators apply worldwide, while research guidance allows for context-sensitive interpretation.
5
Transparent and reviewable
Methodological choices, indicators, and evidence rules are clearly defined so results can be understood, scrutinized, and improved.
What GIRAI Measures
GIRAI evaluates national AI ecosystems across five interconnected dimensions
1
Inclusion and Diversity
How countries govern for equity and inclusion in AI, addressing bias and discrimination, protecting marginalised groups, and enabling their meaningful participation.
2
Ethics and Sustainability
Safeguards against discrimination and harm, transparency and explainability requirements, human oversight, and environmental responsibility.
3
Labour and Skills
How countries are preparing workers for an AI-driven economy and protect labour rights, ensuring the benefits of AI are shared broadly across society.
4
Trust and Safety
Data protection and privacy, safety and security frameworks, access to redress, impact assessments, and responses to AI-enabled misinformation or violence.
5
Use of AI in Public Sector Delivery
How governments deploy AI in public services, ensuring its use remains transparent, accountable, and respectful of fundamental rights and democratic values.
Framework Evolution
Why the Framework Evolved
Since the first edition of GIRAI, the global AI governance environment has evolved significantly. Governments have moved from high-level strategy development toward implementation and regulatory experimentation. Oversight mechanisms have matured, public debate has deepened, and expectations around responsible AI have become more precise.
To remain analytically relevant, the GIRAI framework was refined to better capture these developments. The evolution reflects lessons learned from the previous cycle, feedback from researchers and stakeholders, and the need to distinguish more clearly between policy intent, operational action, and structural capacity.
“The core principles of GIRAI remain unchanged. What has evolved is the precision with which those principles are measured.”
GIRAI Methodology — Second Edition
How the Framework Was Refined
Strengthening Clarity, Comparability, and Implementation Focus
Clearer Dimension Structure
Dimensions were refined to better distinguish between governance commitments, implementation activity, civil society contributions, and enabling conditions.
Stronger Indicator Definitions
Indicator wording and evidence requirements were tightened to reduce ambiguity and improve consistency across countries, enhancing cross-country comparability.
Greater Emphasis on Implementation
The updated framework places stronger focus on operational oversight, enforcement, and real-world action, not just the existence of policies on paper.
Enhanced Review and Validation
Quality assurance processes were strengthened, including clearer coding guidance, multistage reviews, a scientific advisory committee, a global forum for peer consultation among country researchers, and more structured cross-country validation procedures.
What Changed in This Edition
Key differences between the previous and current edition of the GIRAI methodology framework.
DimensionPrevious EditionCurrent Edition
Governance Emphasis
Previous EditionBefore
Greater emphasis on the existence of frameworks
Current EditionNow
Clear separation between government frameworks, implementation actions, and contextual indicators
Policy vs. Implementation
Previous EditionBefore
Less distinction between policy intent and implementation
Current EditionNow
Stronger focus on operational oversight and enforcement mechanisms
Indicator Structure
Previous EditionBefore
Broader grouping of indicator types
Current EditionNow
Refined indicator definitions and clearer evidence standards
Governance Context
Previous EditionBefore
Early-stage global governance landscape
Current EditionNow
Expanded validation and review procedures reflecting a maturing governance environment
Key Terms and Definitions
The core concepts that shape GIRAI's methodology, clearly defined for transparency and consistency.
Responsible AI issues through which AI governance frameworks, government actions, civil society engagement and government deployment of AI, are all assessed.
Components used to measure the extent to which government frameworks fulfil the requirements of a specific AI policy indicator. Only the 17 AI Policy indicators have thematic breadth.
One of five areas of responsible AI governance into which GIRAI groups its indicators, namely Inclusion and Diversity, Ethics and Sustainability, Labour and Skills, Trust and Safety, and Use of AI in Public Service Delivery.
One of three groupings that organise all GIRAI indicators — excluding the "Unacceptable Risks AI Systems" indicator — by type: AI Policy (17 indicators, assessed through primary data), Civil Society Engagement (5 indicators, assessed through primary data), and Enabling Conditions (15 indicators, assessed through secondary data).
The laws, policies, strategies, and institutional arrangements through which governments commit to governing AI responsibly.
Operational measures taken by the government to implement responsible AI — including enforcement, oversight, funding mechanisms, knowledge generation, and capacity-building programmes.
Initiatives and contributions from civil society organisations that shape, scrutinize, and advance responsible AI.
GIRAI Evidence Standards
To ensure credibility across 135 countries and jurisdictions, GIRAI applies strict evidence rules:
1
Publicly available and verifiable
All evidence must be accessible for independent review.
2
Written documentation only
Ensures consistency and permanence of evidence.
3
Archived online sources
Web sources are permanently archived for future retrieval.
4
Within study timeframe
All evidence must be within the specified timeframe of the GIRAI second edition.
5
AI-related explicitly
All evidence must directly relate to areas of responsible AI.
6
No standalone interviews
Interviews inform but don't constitute evidence
OPEN DATAAccess the Data
Access the Data
Behind the Index
Download the full dataset and explore the structured evidence, indicators, and documentation that underpin GIRAI scores across countries.
Research and Review Process
How multi-layered review ensures consistency, independence, and methodological rigor.
150+
Researchers engaged globally
3-Tier
Independent review structure
25
Indicators assessed
100%
Public evidence verified
Research Process
Five Stages of Evidence Review
From in-country fieldwork to global validation, each submission advances through structured, independent checkpoints.
1
Expert-Led Research
In-country researchers collect and document evidence using standardised questionnaires and methodological guidance.
2
Researcher Submission
Completed questionnaires and supporting evidence are formally submitted through the survey system for review.
3
Country Coordinator Review
Country coordinators review submissions, verify evidence quality, and ensure alignment with methodological requirements.
4
Regional Supervision
Regional supervisors conduct oversight, spot-check submissions, and address inconsistencies across countries.
5
Global Review
Final global review and validation to ensure ensure consistency, accuracy, and cross-country comparability.
Download GIRAI Research
Methodology Handbook
Access the full technical documentation behind the GIRAI methodology.