Responsible AI Guide for Public Administration

More than a technical guide, this resource establishes principles, guidance, and practical instruments to support the implementation of responsible AI projects, aligned with public service values and the European regulatory framework.

What the Responsible AI Guide is

Developed by the Agency for Technological Reform of the State (ARTE), in coordination with the Council for Information and Communication Technologies in Public Administration (CTIC), the Responsible AI Guide was designed to support public bodies in making informed decisions about Artificial Intelligence technologies.

The document presents a practical approach to AI governance, promoting principles such as the protection of fundamental rights, human oversight, risk mitigation, and accountability. At the same time, it seeks to foster a culture of responsible innovation, reinforcing citizens' trust in digital public services.

Who it is for

The guide is aimed at professionals and teams within Public Administration involved in digital transformation processes, including:

• Leaders and heads of innovation;

• Project managers;

• Technical and development teams;

• Data and technology specialists;

• Decision-makers involved in the procurement or implementation of digital solutions.

The five pillars of responsible AI

The methodology presented in Mosaico is organised around five fundamental dimensions of responsible AI:

• Accountability: ensuring that clear governance roles, processes, and mechanisms are in place;

• Transparency: ensuring that systems and decisions can be understood and audited;

• Explainability: enabling the interpretation of results and the justification of automated decisions;

• Fairness: preventing bias and promoting equitable treatment;

• Ethics: aligning the AI use with human rights and public service values.

These pillars serve as the basis for the assessment, design, and monitoring of AI projects in the public context.

Ethical Risk Assessment Tool

Complementing the guide, Mosaico also provides an Ethical Risk Assessment Tool, structured as a checklist. This tool enables teams to assess risks throughout the entire lifecycle of a project — from conception through to implementation and monitoring — translating ethical principles into concrete analysis and decision-making criteria. The aim is to support the early identification of risks, document technical choices, and strengthen compliance and institutional trust.

Alignment with the European framework

The Guide also serves as a preparation tool for the AI Act application, the European regulation on Artificial Intelligence. In force since 2024, this framework establishes different risk levels for AI systems — from minimal risk to high-risk applications — imposing specific obligations in areas such as technical documentation, human oversight, transparency, and risk management. For public bodies using AI in sensitive areas such as human resources, social benefits, or enforcement, this alignment is particularly relevant.

How to access

The Responsible AI Guide and the Ethical Risk Assessment Tool are available on the Mosaico portal, in the area dedicated to Artificial Intelligence.