Summary: UNESCO’s Perspective on AI and Ethics

UNESCO’s Recommendation on the Ethics of Artificial Intelligence is the first global standard-setting framework for guiding how AI should be developed and used. It was adopted by all 193 UNESCO Member States in 2021.


Core Values

UNESCO emphasizes that AI must respect and promote:

  • Human rights and dignity – AI should never undermine basic freedoms.
  • Human oversight – people must remain responsible for decisions, not machines.
  • Inclusion and fairness – AI should work for everyone, avoiding discrimination or bias.
  • Transparency and explainability – AI processes and decisions must be clear and understandable.
  • Sustainability – AI should support environmental and social wellbeing, not harm it.

Key Ethical Principles

  1. Proportionality & Do No Harm – Use AI only when it provides clear benefits and avoid risks of harm. 
  2. Safety & Security – Protect people from technical failures, misuse, or malicious attacks.
  3. Privacy & Data Protection – Safeguard personal data and ensure informed consent.
  4. Fairness & Non-Discrimination – Prevent bias based on gender, race, disability, or status.
  5. Accountability & Responsibility – Establish clear roles so institutions and individuals can be held accountable.
  6. Awareness & Literacy – Build understanding of AI among citizens, policymakers, and educators.


Policy Action Areas

UNESCO recommends governments and institutions focus on:

  • Ethical impact assessments before deploying AI.
  • Readiness assessments to check institutional capacity.
  • Strong governance frameworks (laws, policies, oversight bodies).
  • Capacity building for officials, teachers, and communities.
  • Environmental sustainability – minimizing AI’s carbon and resource footprint.

Why It Matters

For education and humanitarian contexts (like Education in Emergencies), UNESCO’s framework reminds officials that AI tools handling sensitive learner data must:

  • Protect privacy and security of children and displaced populations.
  • Ensure data and models are not biased against vulnerable groups.
  • Keep human judgment central in all high-stakes decisions.
Last modified: Friday, 26 September 2025, 12:31 PM