The White House is preparing an AI cybersecurity framework under Executive Order 14409 that includes a classified benchmarking process for covered frontier models (advanced AI systems). The process will be led by the National Security Agency, with the NSA Director holding final authority over designations. CISA, the Treasury Department and NIST will also contribute. Companies may voluntarily seek government access for review, but the benchmark’s details have not been disclosed. The framework could create a more formal evaluation process and potentially favor established developers such as Anthropic. Market pricing suggests participants see a possibility that Anthropic’s AI model will be ranked best by September 2026. Markets are expected to track further details, the framework’s effects on model evaluations and how Anthropic and its competitors respond to the NSA’s role.