Artificial intelligence (AI) systems may be used to improve access to social security benefits, getting the right benefits to the right people at the right time. Adopting AI technology may also complement broader public sector efforts to enhance productivity and deliver more user-centred and proactive services.
To explore how AI can be used to this end, the main social security institutions of France (the Caisse nationale des allocations familiales – CNAF, the Caisse nationale de l’assurance maladie – CNAM, and the Mutualité Sociale Agricole – MSA) and Italy (the Istituto nazionale della previdenza sociale – INPS) have engaged in a collaborative project, funded by the European Union via the Technical Support Instrument, and implemented by the OECD, in co-operation with the European Commission.
The institutions undertaking this project are already testing AI applications in a variety of functions, such as enhancing internal knowledge management, improving communication with existing and potential beneficiaries, and supporting data analysis, while still proceeding cautiously given the high-risk designation under the EU AI Act for many use cases.
The purpose of this report is to provide a curated set of international good practices to guide future AI adoption, based on gaps and challenges identified in the earlier gap analysis report developed by the OECD. The report also documents concrete policy levers, tools, implementation strategies, and capacity-building efforts relevant to public service contexts at the national level, with particular attention to issues of data quality, governance, and workforce readiness. It aims to provide a guide on how the social security sector can benefit from and align with national efforts for more cohesive, impactful and trustworthy uses of AI. Further efforts will be needed to explore the effective use of AI in the social security sector, particularly in adapting these practices to the specific operational, legal, and ethical contexts of social protection systems. Continued experimentation, evaluation, and cross-sector collaboration will be essential to ensure that AI adoption delivers tangible benefits while safeguarding equity, transparency, and public trust.
Ultimately, this work underscores a critical opportunity to harness AI as a strategic enabler of more inclusive, effective, and transparent social security systems. Realising this potential will depend on embedding robust governance frameworks, investing in data and infrastructure, and building the institutional and human capabilities necessary for trustworthy AI in the public sector.