newsnews.ai

Trump signs executive order seeking early government access to AI models

The new directive establishes a voluntary framework for federal agencies to review powerful AI systems before their public release to address national security risks.

By NewsNews AI
The White House, Washington, D.C. USA
The White House, Washington, D.C. USA·Photo: Diego Delso via Wikimedia Commonscc-by-sa

New Framework for Model Review

President Donald Trump signed an executive order on Tuesday that requests artificial intelligence developers provide the federal government with early access to "frontier models". The order establishes a voluntary framework intended to allow the government to vet and test powerful AI systems for safety issues before they are deployed to the public.

The directive was signed in private. Under the new rules, tech companies will be asked to share their models with the government for review on a voluntary basis. This approach asks companies to keep the government in the loop about their work, while leaving the decision to participate at the discretion of the developers.

National Security and Cybersecurity Focus

The executive order is described as an effort by the president to address national security threats and tighten cybersecurity defenses. Specifically, the order directs several federal agencies—including the Pentagon, the Treasury Department, and the Cybersecurity and Infrastructure Security Agency (CISA)—to strengthen the nation's cybersecurity protections for critical infrastructure.

As part of these efforts, the administration aims to give officials the opportunity to brace the economy against potential security risks associated with advanced AI. The order also directs officials to create a classified system designed to evaluate the cybersecurity capabilities of advanced AI models.

Implementation and Benchmarking

The executive order sets a 60-day deadline for several entities to develop a classified benchmarking process. This group includes the Treasury Department, the National Security Agency (NSA), CISA, the National Institute of Standards and Technology (NIST), and White House officials.

This benchmarking process is intended to assess the advanced cyber capabilities of AI models. Through this process, the designated agencies will determine when a specific model should be officially categorized as a "covered frontier model".

Internal Administration Dynamics

The final language of the order reflects internal White House deliberations. Former White House AI czar and current adviser David Sacks, along with National Economic Council deputy director Ryan Baasch, pushed for specific language that prohibits the creation of mandatory government licensing.

Reports indicate a rift within the administration regarding the level of regulation. While some officials sought a more rigid structure, Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent reportedly favored a regulatory framework where companies provide information at their own discretion. This voluntary approach resembles an earlier proposal that President Trump had previously shelved.

Sources (8)Open

Topics

How NewsNews AI made this storyOpen

NewsNews AI researched this story across 8 sources, drafted it, and ran the result through an independent editorial pass. It cleared editorial review on first pass.

  • 8 sources cited · linked in full at the bottom of the article
  • Image license verified · cc-by-sa
  • Independent editorial pass · approved

From the editor

Verified both previously flagged issues are now resolved: the "in the loop" direction correctly reflects source [5] (companies keep the government informed, at their discretion), and "directs" is used instead of "mandates" for the classified system per source [8]. All key claims are supported by their cited snippets, quotes are paraphrased accurately, and no new issues were introduced by the revisions.

More about our editorial process

Feedback

We want to hear from you, especially when something is wrong. No signup, no email required.

Keep reading