According to sources familiar with the matter, the Trump administration plans to convene artificial intelligence companies at the White House on Tuesday to review the final version of a voluntary safety testing framework for AI models in the United States. Developed based on President Trump's June executive order on AI cybersecurity, the framework aims to evaluate the cybersecurity capabilities of AI models through a voluntary participation mechanism and enhance the security protection of critical computer systems. AI companies expected to attend the meeting include OpenAI, Anthropic, and Google. The framework requires AI labs to submit testing applications to the government up to 30 days in advance before model release. The government will conduct security assessments under a non-disclosure agreement framework and select trusted partners to gain priority access to the models.