Trump Administration Seeks Structured AI Model Evaluation with Anthropic, OpenAI

The US government is making progress towards finalizing an AI model evaluation plan, with Anthropic and OpenAI working closely with the Trump administration to review frontier AI models for cybersecurity and national-security risks.

This plan is part of a broader initiative directed by a recent executive order, which mandates the creation of a classified benchmarking process and a framework that allows developers to grant the government access to advanced models up to 30 days before their release.

The collaboration between Anthropic, OpenAI, and the Trump administration indicates a shift from informal consultations to a more structured evaluation process, with both companies adjusting their existing pre-release review agreements with the Commerce Department to align with the new AI Action Plan.