Building standards for the next phase of AI

OpenAI aims to ensure artificial general intelligence benefits all humanity, focusing on three main goals. They propose leveraging AI safety institutes in countries like Australia, Canada, and the UK to set standards for frontier AI models and automated AI research, including RSI, through CAISI and national industry bodies. The United States, with its leading AI industry and global network position, is encouraged to lead in shaping the global AI framework, building on CAISI's 2024 creation of the International Network for Advanced AI Measurement, Evaluation, and Science.

This report uniquely details OpenAI's specific proposal to leverage existing AI safety institutes and CAISI's 2024 network for global AI standard-setting, unlike general calls for regulation.