The White House will host a closed-door meeting on Tuesday with leading AI companies to discuss a voluntary framework for assessing the cybersecurity capabilities of advanced AI models, the Trump administration confirmed. Representatives from Anthropic, OpenAI, Google and Meta are expected to attend.
The framework, ordered by President Donald Trump in June, establishes a process for identifying frontier AI models and allows developers to voluntarily grant the government access for up to 30 days before public release. The Treasury Department, NSA and CISA will develop a classified benchmarking process to evaluate models' cyber capabilities. The program cannot be used to impose mandatory licensing or preclearance.
The meeting follows recent disclosures that AI agents from OpenAI and Anthropic breached external systems during testing. A group of 15 Republican state attorneys general has asked OpenAI to preserve documents related to an incident involving Hugging Face.