Background
On June 2, 2026, President Donald Trump signed Executive Order 14409, titled "Promoting Advanced Artificial Intelligence Innovation and Security." <cite index="10-3,10-4">The order reflects the administration's stated policy of advancing U.S. leadership in artificial intelligence (AI) while addressing national security risks associated with increasingly capable AI systems.</cite> <cite index="11-5">To accomplish these policy goals, the order outlines two approaches: strengthening U.S. government and private industry cyber defenses in response to advanced AI, and developing voluntary benchmarking and review frameworks for secure development and release of frontier AI models.</cite>
The 30-Day Review Mechanism
<cite index="9-3">The order directs the National Security Agency (NSA) to classify which AI models are "covered frontier models" via a secret benchmarking process, allows developers to voluntarily give the government 30 days of early access before broader release, and establishes a government-industry AI cybersecurity clearinghouse.</cite> <cite index="16-3">The order contemplates federal access to covered frontier models for up to 30 days before release to other trusted partners, subject to confidentiality, cybersecurity, insider-risk, and intellectual property (IP) protections.</cite>
<cite index="2-9,2-10">The 30-day period is a maximum access window, not an automatic delay imposed on every model. The order does not state that companies must wait the full month, nor does it require a model to remain unreleased if federal evaluators have not completed their work.</cite> Critically, <cite index="9-2">the order explicitly prohibits agencies from treating the framework as a mandatory licensing, permitting, or preclearance requirement.</cite>
Participating Labs and the August 1 Deadline
<cite index="19-3">The White House is in advanced talks with Anthropic, OpenAI, Google, Microsoft, and Amazon to finalize voluntary standards governing how the most capable AI models are tested and released, according to a Financial Times report published July 2.</cite> <cite index="20-1">August 1 marks 60 days from the executive order's signing—the deadline by which the NSA must finalize a classified benchmarking process for designating covered frontier models and by which the multi-agency group must publish a formal voluntary framework governing how that review process works.</cite>
<cite index="1-1">OpenAI and Anthropic signed comparable Center for AI Standards and Innovation (CAISI) agreements in 2024; Google DeepMind, Microsoft, and xAI joined in May 2026.</cite> Meta remains a notable absentee: <cite index="1-7">the review structure now spans its major peers but not Meta, and that gap has reportedly drawn direct White House attention.</cite>
Jailbreak Scoring and Clearinghouse
Embedded within the broader framework is a parallel security initiative. <cite index="23-4,23-5">Anthropic proposed a consensus industry framework for scoring the severity of AI jailbreaks, explicitly modeled on the Common Vulnerability Scoring System (CVSS) used in traditional software security, and is drafting it alongside Amazon, Microsoft, Google, and other partners.</cite> <cite index="5-3,5-4">The core mechanism also includes a cybersecurity clearinghouse, driven by collaboration among key federal agencies including the Treasury Department and the National Security Agency.</cite>
Regulatory Implications
The framework's voluntary design does not eliminate practical pressure on frontier labs. <cite index="6-1">What the GPT-5.6 episode established, regardless of what August 1 produces, is the operating precedent: frontier AI models in the United States are now subject to government review before public deployment, de facto if not de jure.</cite> <cite index="13-2">The order emphasizes collaboration with industry and avoids imposing mandatory regulatory requirements, with its ultimate impact expected to depend on subsequent agency implementation.</cite>
<cite index="15-4">The executive order follows a series of administration actions promoting U.S. AI leadership, including the January 2025 EO "Removing Barriers to American Leadership in Artificial Intelligence" and the December 2025 EO "Ensuring a National Policy Framework for Artificial Intelligence."</cite> Legal analysts at Skadden advise that frontier model developers should <cite index="15-1,15-2,15-3">be prepared to evaluate whether their models meet the forthcoming covered frontier model designation, prepare for the classified benchmarking process, and consider whether product release timelines should account for a potential government access period of up to 30 days.</cite>