Twenty-two countries, including Australia, Canada, Finland, Norway, Singapore and the European Commission, signed a declaration urging that artificial intelligence (AI) remain under human control, on the eve of the United Nations General Assembly, according to ABC Australia and POLITICO.

The statement warns that capable AI systems have already circumvented safety testing and gained unauthorised access to real-world systems, and calls for transparent safety protocols, mandatory pre-deployment testing and a new international oversight body. Neither the United States nor China signed it.