U.S. executive order narrows AI model vetting framework after White House dispute

U.S. executive order narrows AI model vetting framework after White House dispute
US narrows AI vetting

The Trump administration is introducing a voluntary process to secure early government access to advanced AI models as officials weigh cyber security risks tied to frontier systems. The order reduces the proposed review window to 30 days from an earlier 90-day draft, underscoring a compromise between security concerns and pressure to avoid slowing U.S. AI developers.

Highlights

  • Donald Trump's executive order requires AI labs to submit new frontier models for security reviews up to 30 days before public release.
  • The revised order, scaled back from an earlier draft, asks AI companies to select trusted partners for early access and cybersecurity support, balancing oversight with innovation.
  • Urgency behind the measure stems from Anthropic’s forthcoming Mythos model, which officials say could identify and exploit cybersecurity vulnerabilities, prompting pre-release access for U.S. agencies.

Revised order sets shorter review period

As first reported by Financial Times, Donald Trump signs an executive order that asks AI labs to submit new frontier models for security reviews for up to 30 days before public release.

The measure is a scaled-back version of an earlier draft that was expected last month but was withdrawn after concerns inside the administration that tighter oversight could hinder the expansion of major U.S. AI groups including OpenAI, Google and Anthropic. Tuesday’s order also asks AI companies to work with the federal government to choose trusted partners that can receive early access to covered frontier models to support secure innovation and strengthen the cyber security of critical infrastructure.

The push gains urgency from concerns over Anthropic’s forthcoming Mythos model, which officials say is capable of identifying and exploiting cyber security vulnerabilities. U.S. agencies are seeking access before the model becomes publicly available, while governments and companies in other countries are also seeking early access to help patch weaknesses the system may uncover.

Security concerns collide with innovation agenda

Administration officials have signaled that the government may rely on the cyber expertise of large banks, financial institutions and specialists in other sectors to help assess advanced models. That approach reflects an effort to spread vetting across industries that could face direct exposure to the risks and benefits of frontier AI deployment.

The order arrives amid continuing disagreement within Trump’s orbit over how far Washington should go in overseeing advanced AI. Some allies favor bringing leading models under stronger government control, while others warn that aggressive intervention could weaken U.S. competitiveness in the global AI race.

Kevin Hassett, director of the National Economic Council, at one stage argues that frontier models should face an approval process similar to pharmaceuticals and be released only after they are proven safe. The idea draws strong opposition from AI founders and investors close to the administration, and Treasury secretary Scott Bessent says the White House does not view the order as a new regulatory burden, stressing that the administration wants to keep security in focus while preserving U.S. leadership in AI.

Our earlier coverage of Microsoft’s Q3 FY2026 update highlighted surging AI investment and revenues alongside rising regulatory scrutiny in the U.S., which continues to shape sentiment around Big Tech’s AI ambitions. We noted Microsoft’s heavy capex plans, Copilot platform changes, and the way oversight pressure can add volatility even when fundamentals look strong.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.