“Major AI companies are asking governments to create international agreements for a level playing field on safety, requiring all companies to demonstrate their systems are not dangerous before deployment.”