Artificial intelligence in public services and gender gaps: the figures from the 2026 OECD Forum

The Organisation for Economic Co-operation and Development (OECD) devoted the second edition of its Forum on Gender Equality to digital transformation. The background paper gathers the evidence on how artificial intelligence is being used in public services and on which gaps remain open.

The Forum and the background paper

The OECD Forum on Gender Equality took place on 8 and 9 July 2026 at the OECD Conference Centre in Paris, under the theme «Harnessing the Digital Transformation for All». It was the second edition, after the inaugural one in 2024, and brought together policymakers, experts and practitioners around the relationship between digital technology and gender equality.

The background paper, the Key Issues Paper published by the OECD in 2026, is organised in seven chapters: artificial intelligence and public services; digital technologies, data and the care economy; skills and leadership; women’s entrepreneurship; gender-based violence online against women in public life; children’s safety in the digital environment from a gender perspective; restrictive masculinity norms in online spaces. This note focuses on the first chapter.

The use of artificial intelligence in public services

According to the OECD report Governing with Artificial Intelligence, published in June 2025 on the basis of two hundred use cases across eleven government functions, 67% of OECD countries use artificial intelligence to improve public service design and delivery. The areas identified as high impact are health, justice, social protection and employment services; around half of Public Employment Services in OECD countries use artificial intelligence tools to match jobseekers with vacancies and to support job-search strategies.

The Forum paper notes that these tools can reduce some of the barriers that particularly affect women: care responsibilities, geographic isolation, informal employment and limited access to legal or financial services. Examples given include automatic enrolment in benefits and the use of predictive analytics, alongside automated assistance for women-led enterprises in public procurement.

The gaps recorded in the data

The statistical picture set out in the paper shows differences both in the relationship with services and in access to skills. The 2022 OECD Risks that Matter survey finds that only 19% of women believed applications for public benefits would be simple and quick, compared with 25% of men, and that 50% of women reported difficulties in accessing benefits, compared with 43% of men.

On tools, 29% of the women surveyed report using artificial intelligence tools at work, compared with 41% of men, and among young people aged 16 to 24 more than twice as many men as women report being able to programme. The OECD report Gender Equality in a Changing World (2025) adds that in 2023 women accounted for between 11% and 24% of information and communication technology specialists in OECD countries; that over the last two decades an average of 6% of venture-capital-funded digital start-ups were founded by women only and 15% by at least one woman; and that 11% of scientific papers on artificial intelligence have female authors only. The paper also notes that women are over-represented in clerical and administrative roles most exposed to automation.

The risks identified for system design

The chapter sets out three orders of risk. The first concerns data: many systems rely on administrative datasets that reflect pre-existing inequalities in access to services, employment or income, with possible effects on access to social benefits, healthcare prioritisation, creditworthiness assessment and job matching. The second concerns access: the benefits of automated services do not reach those without an internet connection or digital skills. The third concerns the composition of the teams designing the systems, where limited female presence makes it less likely that certain needs are reflected.

The paper also records capacity constraints within public administrations: limited technical expertise, fragmented governance arrangements and insufficient integration of gender considerations into digital strategies, with the resulting difficulty in detecting discriminatory outcomes, providing redress and assigning responsibility when harm occurs.

Governance recommendations and the European framework

Five recommendations are addressed to governments: embedding gender equality objectives in artificial intelligence strategies and service design frameworks; strengthening the governance of gender-disaggregated data while upholding privacy and data protection standards; setting transparency, explainability and traceability requirements for the tools used by administrations; investing in technical competencies alongside expertise in gender analysis, ethics and public governance; and providing participatory mechanisms involving users and civil society organisations. The international instrument cited is the OECD Recommendation of the Council on Artificial Intelligence (OECD/LEGAL/0449). Among the national practices recalled, Canada has adopted a mandatory algorithmic impact assessment tool for automated decision systems.

In the European Union the same area is covered by Regulation (EU) 2024/1689 on artificial intelligence, whose Annex III classifies as high risk the systems intended to determine access to essential public services and public assistance benefits; Article 27 requires a fundamental rights impact assessment from deployers that are bodies governed by public law or private entities providing public services. The application timeline of the Regulation is the subject of the Digital Omnibus proposal, whose legislative process is not concluded.

Outlook

The Forum documents, the Key Issues Paper and the co-chairs’ summary, remain the reference for OECD work on gender and digital technology; the feedback survey on the 2026 edition is open until 10 September 2026. It remains to be defined which national strategies will take up the recommendations and by which indicators the effect will be measured, since the paper sets neither quantitative targets nor deadlines. On the European side, the application of the rules on the high-risk systems listed in Annex III depends on the outcome of the legislative process under way.

In dialogue with the 2030 Agenda

Goal 5 (Gender equality). The axis of the paper: access to skills, services and decision-making roles in the digital transformation.
Goal 16 (Peace, justice and strong institutions). Transparency, accountability and redress in the automated tools used by administrations.
Goal 9 (Industry, innovation and infrastructure). The composition of the teams developing the systems and access to venture capital.
Goal 10 (Reduced inequalities). The benefits of automated services do not reach those without connectivity or digital skills.
Goal 8 (Decent work and economic growth). Exposure of clerical roles to automation and reskilling policies.