UNESCO adopts a global recommendation on AI ethics

All 193 UNESCO member states adopted the Recommendation on the Ethics of Artificial Intelligence at the organisation’s General Conference, the product of roughly two years of drafting and consultation and the first attempt at a normative AI framework agreed by nearly every country in the world at once. UNESCO director-general Audrey Azoulay called it “the first global normative framework” for AI, while leaving states responsible for applying it domestically.

Like other UNESCO recommendations, the text is non-binding: it creates no enforcement mechanism and no penalty for non-compliance, only a set of values, principles and policy actions that signatory states commit to pursue. Its substantive content leaned further than many national AI policies of the time — it explicitly calls for member states to ban the use of AI systems for social scoring and mass surveillance, and sets out policy areas covering data governance, environmental sustainability, gender equality, labour and healthcare, all framed against existing human-rights instruments and the UN Sustainable Development Goals.

As soft law, the recommendation’s practical effect depended entirely on individual governments choosing to legislate in its spirit, and enforcement questions dominated criticism of it in the years that followed, particularly given several signatory states’ own use of surveillance and scoring systems the text nominally opposed. It nonetheless stood, for several years, as the broadest international consensus document on AI governance, predating the EU’s binding AI Act and most national AI legislation by two or more years.