Recently, renowned artificial intelligence research institute OpenAI and Anthropic announced a cooperation agreement that allows the U.S. government to have early access and evaluation of new AI models being developed before they are launched on the market. This initiative aims to further enhance the security of AI technology, help identify potential vulnerabilities, and mitigate the risk of negative impacts. The AI Safety Institute welcomes this move, seeing it as significant for assessing the safety risks of AI models and reducing potential issues. This new development under the collaboration highlights the trend of cooperation between tech companies and governments in promoting innovation and secure application of AI technology.