In an About-Face, Trump Administration Considers Vetting New AI Models

If it happens, the move would mark a stark reversal for the administration, which up to now has pursued a largely laissez-faire, anti-regulatory approach to the technology. Trump signed an executive order shortly after returning to office overturning a prior EO issued by President Biden that had established government-wide regulations of AI. In December, Trump issued another EO seeking to eliminate what it called “unduly burdensome” AI regulations including establishing a federal policy to preempt state-level AI statutes.

The proposed new working group is likely to consider a number of oversight approaches, the Times’ sources said. But a review process could be similar to one being developed in the U.K., which has assigned several government bodies to ensure that A.I. models meet certain safety standards.

According to the Wall Street Journal, the administration’s discussions were triggered by the release of Anthropic’s Mythos model and other powerful AI systems. The goal is to protect consumers and businesses from cyberattacks and other disruptions caused by the premature release of such models, the Journal’s sources said.

In recent weeks, National Cyber Director Sean Cairncross has held meetings with other administration officials and tech-industry leaders as part of the administration’s response to Mythos. Vice President JD Vance and Treasury Secretary Scott Bessent have been in meetings on the subject, per the Journal, and Bessent has warned financial-industry executives about the risks such models could pose.

Related: Bessent, Powell Summon Bank CEOs Over Anthropic’s New AI Model  

The Trump administration’s recalibration comes as the public, as well as lawmakers on Capitol Hill, are increasingly skeptical of the unregulated growth of AI and its impact on jobs, energy prices, privacy, health care and any number of other sectors of the economy. A Pew Research poll released in November found broad bi-partisan unease, with 50% of Republican voters and 51% of Democrats expressing more concern than excitement over the technology. An NBC News survey released last month found nearly twice as many consumers hold negative views of AI than positive ones.

Resistance to AI data centers has also emerged as hot political topic heading into the fall’s midterm elections, as communities slated to host data centers press local and state officials to stem the tide amid increasing energy costs and concern over their environmental impact. According to Data Center Watch, $156 billion worth of planned data center development was blocked in 2025 due to local opposition.

The administration’s desire to preempt state AI regulations also has not deterred state governments from moving ahead on new technology restrictions, as seen most recently in Connecticut, where an online safety bill that would regulate AI passed both statehouse chambers with overwhelming bipartisan majorities, according to CT New Junkies.

Among the tasks being considered for the administration’s proposed new AI working group would be helping determine which agencies should be involved in vetting new models, according to the Times’ sources. The administration is also considering reviving the Center for AI Standards and Innovation, an agency set up by the Biden administration to vet AI models that was been largely sidelined by Trump’s initial AI executive order.