U.S. Secretary of Commerce Raimondo stated that the U.S. Artificial Intelligence Safety Institute is collaborating with OpenAI and Anthropic to conduct voluntary pre-deployment testing of AI models, and is working with the UK Safety Institute to release the first joint government pre-deployment test of advanced AI models. The institute aims to support industry safety efforts without hindering progress. The organization emphasizes ensuring that AI serves humanity and addresses the threat of widespread unemployment. Raimondo has previously stated that AI technology is unprecedented in history and has significant potential to disrupt the job market. Governments must ensure the safety of AI systems before they are released, emphasizing the need for rigorous testing and evaluation. Establishing international best practices for AI safety is a matter of urgency. Over the past year, progress has been made in promoting AI safety, and relevant initiatives have been established that did not exist before. The International AI Safety Institute Network currently has 10 members, with plans to add more in the future. The U.S. Department of Commerce's AI Safety Institute focuses on identifying national security and public safety risks associated with AI. A team of scientists and engineers is developing advanced methods for testing before and after the deployment of AI models. The institute emphasizes scientific excellence rather than acting as a regulatory body. Its goal is to ensure that the government’s AI work is both intelligent and cutting-edge, guided by science. (Jinshi)
