Andorran Code of Ethics for Artificial Intelligence

May 2024

View the original document

Strategies and Action Plans

The Andorran Code of Ethics for AI is both a guiding tool and a preparatory step for binding regulation. It positions Andorra within the international ethical AI movement while reflecting its own digital transformation agenda. Its emphasis is on human rights, inclusion, sustainability, and trustworthiness, making it a holistic reference for AI development, deployment, and governance across all sectors.

Introduction and context

The Code acknowledges that AI is reshaping society, the economy, and governance. It stresses that AI brings both opportunities (innovation, competitiveness, efficiency) and risks (to rights, freedoms, and trust). Andorra situates its approach within global initiatives, citing UNESCO’s Recommendation on AI Ethics (2021), UN resolutions on safe AI (2024), OECD guidelines, the Council of Europe’s AI treaty process, and the EU AI Act. It also ties the Code to Andorra’s Digital Transformation Programme (PdTDA) 2024–2027, which has four strategic axes: digitalising public administration, supporting private sector transformation, consolidating new technologies, and safeguarding digital rights.

Scope and objectives

The Code applies broadly to public authorities, private sector entities, and citizens, including education and research. It is a voluntary guide but may become binding in public procurement contexts. Its aims are to:

  • Support people-centred AI that protects dignity, autonomy, and rights.
  • Help both sectors implement trustworthy AI throughout system lifecycles.
  • Provide foundations for Andorra’s future AI regulation.
  • Promote digital inclusion, transparency, awareness, and skills development.

Values and principles

The Code sets out core values:

  • People-centred digital transformation, ensuring AI serves humanity.
  • Legal and ethical design from the outset (“Human Rights by Design”).
  • Respect for human dignity and rights, especially for vulnerable groups.
  • Diversity, inclusion, sustainability, and environmental protection.
  • Prevention of harm, manipulation, or discrimination.
  • Peaceful and fair application of AI.

And it articulates key principles:

  • Inclusive and sustainable development.
  • Human-centred approach and equity.
  • Necessity, proportionality, and harmlessness of AI.
  • Robustness, safety, and prevention of damage.
  • Transparency, explainability, traceability, and accountability.
  • Privacy and data protection.
  • Human oversight and responsibility.

Trustworthy AI framework

The Code outlines requirements for trustworthy AI:

  1. Human action and oversight.
  2. Technical robustness and security.
  3. Data protection management.
  4. Transparency and explainability.
  5. Diversity, non-discrimination and equality.
  6. Social and environmental well-being.
  7. Accountability and auditability.

It also lists methods (technical and non-technical) such as quality datasets, explainability tools, impact assessments, stakeholder participation, audits, education, and compliance testing. It adopts the ‘Human Rights by Design (HRbD)’ principle, recommending human rights impact assessments for AI systems.

Specific themes

  • AI and ESG: AI should support sustainability, climate goals, and social inclusion, while avoiding excessive environmental footprints and human rights risks.
  • Digital rights and neurorights: The Code highlights the need to protect mental privacy and cognitive integrity against misuse of AI and neurotech.
  • Generative AI: Recognises opportunities in creativity and education but warns about disinformation, copyright, and governance risks. Calls for transparency, traceability, and rights-respecting use.

Fields of action

The Code provides tailored recommendations for:

  • Public sector: transparent, rights-based AI use; training civil servants; promoting open data; ensuring oversight; preserving Andorra’s linguistic and cultural diversity; engaging in international AI governance.
  • Private sector: adopting responsible AI models, preventing human rights abuses, applying ESG criteria, and ensuring accountability in supply chains.
  • Citizens: awareness, literacy, and inclusion in shaping AI policies.