White House Frontier-AI Review Framework Due to Be Finalized by August 1

A June 2026 US executive order established a voluntary framework letting frontier-AI developers give the federal government pre-release access to review a new model’s national-security implications for up to 30 days before wider release — down from a 90-day window in earlier drafts. The framework is due to be finalized around August 1, 2026, and reporting suggests OpenAI, Anthropic and Google are participating while Meta is not; the evaluation benchmarks are classified. Crucially, the order explicitly bars any mandatory licensing or pre-clearance requirement. Paired with the EU AI Act’s August 2 enforcement start, it marks a week when both Washington and Brussels formalized how they intend to look inside frontier models before the public does.