Federated learning for children’s data

Across the world, governments are prioritizing the protection of citizens' data – especially that of children. New laws, dedicated data protection authorities, and digital infrastructure initiatives reflect a growing recognition that data is not just an asset, but a foundation for public trust. 

Yet a major challenge remains: how can governments use sensitive data to improve outcomes – such as in education – without undermining the very privacy protections they are committed to uphold?

One promising answer lies in federated, governance-aware approaches to data use. But realizing this potential requires more than new technology; it demands robust data governance frameworks designed from the outset.

Data governance: The missing link

In many countries, ministries of education, health, and social protection each hold pieces of the puzzle that together could provide a more complete picture of children’s learning and well-being. For example, a child’s school attendance, nutritional status, and family circumstances all shape their ability to thrive, yet these records are kept in separate systems.

Efforts to combine such data often run into legal and technical barriers. Centralized data lakes raise concerns about consent, security, and compliance with privacy laws. In fact, many international standards stress the principle of data minimization – the idea that personal information should not be gathered or combined unnecessarily. 

This is where the right data governance frameworks become essential. Effective governance defines clear rules about how data can be accessed, shared, and used – specifying who has the authority, what purposes are permitted, and how rights are protected. These frameworks make it possible to collaborate with data responsibly, especially when it comes to children.

Federated learning: Governance-by-design in action

Federated learning offers an innovative way to put these governance principles into practice. Unlike traditional AI methods, where all data is collected in a central location, federated learning works by sending algorithms to where the data is already stored – such as school servers or ministry databases. There, the model is trained locally, and only safe, aggregated updates (not raw data) are shared back.

But the true promise of federated learning in the public sector is only unlocked when governance is built into the design from the beginning. This is known as governance-by-design – embedding rules and protections directly into the way data is used, rather than treating them as an afterthought.

How do governance frameworks enable federated learning?

• Clear policies: Define who can participate, for what purpose, and under which conditions. Automated systems enforce these rules.
• Transparency and accountability: Keep auditable records of data use and model training, supporting oversight.
• Retaining control: Each institution maintains ownership of its data but can contribute to shared insights for the public good.

Overcoming practical challenges

While federated learning brings great potential, implementing it across government systems comes with practical hurdles:

• Diverse technology environments: Ministries often use different types of databases and systems. Harmonizing these requires coordination and, sometimes, technical upgrades.  
• Local computing power: Running federated learning can require more processing capability than some organizations currently have, especially in low-resource settings.
• Capacity and skills: Technical staff, data stewards, and regulators all need training in both the technology and the new governance frameworks.

Making it work: privacy-enhancing technologies and public infrastructure

To address these challenges, governments and partners are exploring privacy-enhancing technologies (PETs). These are tools and methods that help analyse data without exposing personal details. For example:

• Differential privacy: Adds “noise” to results so individuals can’t be identified.
• Secure multi-party computation: Allows institutions to collaborate on analysis without sharing raw data.

Just as important is building digital public infrastructure to support federated approaches:

• Data registries: Catalogue available datasets, owners, and usage policies.
• Policy enforcement engines: Automatically check if requests comply with rules.
• Collaboration tools: Provide guidance, templates, and training for ministries, regulators, and researchers.

By making federated learning part of a robust public infrastructure, governments can ensure solutions are sustainable and equitable – not just technical experiments.

Toward collective stewardship of children’s data

Federated, governance-aware learning enables governments to move beyond the false choice between innovation and protection. By embedding strong data governance – clear rules, transparency, and oversight – governments can responsibly harness sensitive data for better outcomes, while upholding obligations under international law, including the Convention on the Rights of the Child. 

But investment in technology alone is not enough. Success also depends on institutional readiness, common standards, and broad participation by all stakeholders – especially those representing and safeguarding children.

Ultimately, the goal is not just smarter algorithms, but better systems of trust: systems that serve children, empower governments, and respect rights.

This future is achievable. But it must be done by design.

UNESCO, through its work on AI governance and digital transformation, is committed to supporting Member States in making this vision a reality – in both high- and low-resource settings alike.

For more, visit UNICEF Innocenti's page on good governance of children's data.