The Global AI Accord: A Declaration on the Governance of Artificial Intelligence as a Global Public Interest, with Explanatory Memorandum

The Global AI Accord is an independent, non-commercial proposal for international cooperation on the governance of frontier artificial intelligence. Issued as a consultation text, it argues that artificial intelligence should be recognised as belonging to the global commons, requiring binding agreement, verification and shared responsibility. Its founding principle is the prohibition of autonomous weapons systems that operate without meaningful human control, together with the exclusion of algorithmic authority from nuclear command and control; its central claim is that verification is the point at which all prior governance proposals fail. Inspired by the institutional design of the Montreal Protocol whilst acknowledging the distinct characteristics of frontier AI, the Accord comprises a Declaration of principles together with an Explanatory Memorandum that develops the legal, historical and policy rationale for each principle. It seeks no signatures. It is issued to invite argument, correction and translation from governments, researchers, industry, civil society and the broader public, towards a Version 2.0 shaped by that consultation. Further information: theglobalaiaccord.org.