About the Lab

The Arab AI Governance Lab is a research and capacity-building lab dedicated to the comparative study of how public authority is exercised over, and through, artificial intelligence in the 22 member states of the League of Arab States. Its work is situated in comparative public administration and regulatory governance, and it treats AI governance as a problem of institutional design, administrative capacity and accountability, not as a problem of technology alone.

Scholarship on AI governance has concentrated on a small number of OECD jurisdictions, and the comparative indicators most often used in policy debate measure the formal adoption of strategies and instruments. Far less is known about the administrative conditions under which formal commitments are translated into practice. The Arab region offers a demanding setting for that question: states that share a language and overlapping legal and administrative traditions differ widely in fiscal resources, state capacity and exposure to conflict. That variation permits structured comparison of institutional choices under very different constraints.

(i) Institutional design. Which configurations of mandates, policy instruments and enforcement arrangements do states adopt to govern AI, and how do these diffuse across the region?

(ii) Administrative capacity. Under what conditions do formal commitments become administrative practice, and where does a gap between institutional form and function persist?

(iii) Pathways. Which combinations of conditions are sufficient for effective governance, given that more than one route can lead to the same outcome?

(iv) AI and society. How does AI reshape language, culture, public communication and the relationship between citizens and the state?

Measurement. The Lab maintains a primary-document coding of AI strategies, bodies, statutes and enforcement arrangements for all 22 states, linked to international panels on readiness, digital government, human capital and cybersecurity, and released as open data with codebook and stated limitations.

Theory. The research programme advances four propositions. Formal adoption and substantive practice can decouple, producing a gap between institutional form and function. Governance outcomes depend on the fit between instrument design and administrative capacity, so that instruments designed beyond available capacity are unlikely to be implemented. Composite indicators themselves act as a channel of isomorphic pressure, rewarding the adoption of visible instruments. Algorithmic administration can concentrate administrative capacity while dispersing accountability across vendors, agencies and legal regimes.

Method. With 22 cases, the Lab uses fuzzy-set Qualitative Comparative Analysis to identify equifinal pathways, and it subjects composite indicators to sensitivity analysis before drawing inferences from rankings.

For governments and regulators, the Lab provides evidence differentiated by stage of regulatory maturity and turns it into capacity building: benchmarking against regional peers, analysis of mandates and coordination arrangements, and training on the fit between planned instruments and implementation capacity. Its work on policy sequencing addresses a recurrent risk in public sector reform, the adoption of advanced instruments before the capacity to administer them exists.

For regional organisations and international partners, the evidence base supports peer learning, the design of shared public goods such as common risk classifications and pooled Arabic language resources, and the alignment of technical assistance with the absorptive capacity of recipient administrations.

Research: peer-reviewed articles and monographs on regulatory governance and algorithmic administration. Open data: a cross-national dataset, an index and a data explorer, free to use under CC BY 4.0. Capacity building: executive programmes, regulator training and institutional development for the public bodies that govern AI.

The coding scheme is being extended to a Global South comparison covering South Asia, selected African states and Latin America.

Evidence first: every figure on this site can be traced to a public source or to the open dataset. Respect: scores and stages describe published frameworks and are not a judgement on national policy choices. Openness: data, methods and limitations are published. Independence: training partnerships do not alter coding decisions.

Regulation; readiness and digital government; research output; infrastructure; culture and communication; governance typology; human capital; cybersecurity. See all areas.

Research collaboration, data questions, corrections and speaking invitations: anis.ben.brik@usi.ch.