Background
Over the past decade, the use of artificial intelligence (AI) systems in publicly funded services has grown. Although this technology may improve service efficiency, accessibility, and responsiveness, without careful design and oversight it may also reproduce social disparities and deepen discrimination against vulnerable populations. As the use of AI systems expands, the need to ensure their alignment with the values of equality and fairness has also grown. Accordingly, JDC commissioned the Myers-JDC-Brookdale Institute to review biases and risks inherent in the use of AI systems in publicly funded services. The review also aimed to identify practical directions for promoting algorithmic fairness by ensuring that decisions made or supported by AI systems do not create or perpetuate unjustified bias, discrimination, or unequal opportunities among specific individuals and population groups. The study was conducted in collaboration with Ben-Gurion University of the Negev.
Objective
Develop a systematic understanding of the challenges and risks related to fairness and equality of opportunity in the use of AI models in publicly funded services and to present a set of practical solutions enabling government ministries and public organizations to plan, develop, and implement AI systems in ways that reduce bias and lower the risk to fairness.
Method
The review is based on information from multiple sources: academic articles from the Israeli and international professional literature; gray literature, including research reports and official documents from government ministries and international organizations; media articles; and three semi-structured in-depth interviews conducted online with AI experts in Israel. The data were collected between March and December 2025, and the interviews were conducted between May and July of that year.
Main Findings and Conclusions
The review maps the biases and risks involved in the development and use of AI systems. The findings indicate that early decisions in the design of such systems shape the boundaries of fairness throughout their development. It is therefore essential to ensure careful development processes and controlled pilots.
At the same time, the findings indicate that promoting fairness cannot rely solely on theoretical principles or on risk assessment. Rather, it is recommended to begin with in-depth examination of the system’s objectives. Based on these objectives, the desired definition of fairness for each system should be established at the outset, alongside the development of dedicated expertise among policymakers that integrates technological, social, and institutional understandings. Combined, these aspects create a comprehensive foundation for the design and responsible use of AI systems in public services.
Beyond their mapping, it is necessary to develop and implement existing tools and practices to address risks of bias and inequality, while adapting them to the way the public system operates. This is essential for protecting vulnerable populations and strengthening public trust. Note also that each system requires a tailored set of solutions, and that no single solution is suitable for all.
To conclude, AI systems can be used to counteract biases and reduce inequality, but this depends on responsible development and implementation that combines technological tools with a strong institutional framework that promotes social justice.
Citing suggestion (APA):
Porzycki, V., Livni Navon, I., & Feinstein, O. (2026). Fairness in publicly funded services: principles, risks, and solutions for implementing algorithmic fairness policy in artificial intelligence systems. An international review. RR-076-26. Myers-JDC-Brookdale Institute.