White House Prepares Executive Order to Vet New AI Models

What Happened

The White House is preparing an executive order that would establish a vetting system for newly released artificial intelligence models, according to Kevin Hassett, a senior administration official. The order, reported by Claims Journal on May 6, 2026, would apply to models including Anthropic PBC's Mythos, and is framed around AI security concerns.

Background

The move comes as the U.S. government has increased scrutiny of large AI systems developed by private companies. Anthropic PBC, the AI safety company behind the Claude family of models, has been developing Mythos as part of its expanding model portfolio. The company, founded in 2021 by former OpenAI researchers, has raised billions of dollars in funding and is among a small number of frontier AI developers operating in the United States.

Executive orders related to AI are not new. The Biden administration issued a broad executive order on AI safety and security in October 2023, which directed federal agencies to develop standards and testing requirements for high-capability AI systems. The Trump administration subsequently revoked that order in early 2025, and has since taken a different regulatory approach focused on reducing barriers to AI development while maintaining national security oversight.

What the Proposed Order Would Do

According to the Claims Journal report, the proposed executive order would create a structured vetting process for new AI models before or upon their public release. Hassett indicated the administration is actively working on the framework, though no signing date was announced and the specific criteria for vetting were not detailed in available reporting.

The reference to Anthropic's Mythos model by name suggests the order is designed to address frontier models, meaning those at the leading edge of capability. It is not yet clear from available reports whether the vetting process would be mandatory, which federal agency would administer it, or what consequences non-compliance might carry.

Context: AI Security as a Policy Priority

Federal interest in AI model security has accelerated alongside rapid advances in model capability. Concerns cited by policymakers and researchers have included the potential for powerful AI systems to assist in the development of biological or chemical weapons, enable large-scale cyberattacks, or generate high-volume disinformation. The National Institute of Standards and Technology has published AI risk management frameworks, and the Department of Homeland Security has convened advisory panels on AI safety.

Separately, Congress has been working on legislation governing AI and digital assets. A separate wire report dated May 7, 2026, noted the White House has set a July 4, 2026, target for passage of the Clarity Act, which relates to cryptocurrency regulation. That deadline underscores the administration's broader push to advance technology-related policy measures within a compressed timeframe.

The MedTech Breakthrough awards announced on the same date, May 7, 2026, highlighted commercial AI deployments in healthcare, reflecting how widely AI systems are now being integrated into regulated industries, a factor that has added urgency to federal discussions about oversight frameworks.

Industry Reaction

No public statements from Anthropic PBC regarding the proposed executive order were included in available wire reports at the time of publication. Other major AI developers, including OpenAI and Google DeepMind, had not issued public comments on the reported order as of the filing of this report.

The AI industry has engaged actively with federal policymakers over the past two years, with several leading companies voluntarily committing to pre-release safety evaluations under agreements facilitated by the White House in 2023.

What Comes Next

The White House has not announced a timeline for finalizing or signing the executive order, and further details on the vetting framework's scope and administration are expected to emerge as the drafting process continues.