White House Weighs Safety Reviews for Frontier AI Models

The White House is considering an executive order that would evaluate artificial intelligence (AI) model’s safety before they are released to the public, National Economic Council Director Kevin Hassett said Wednesday. 

Hassett’s comments follow Anthropic’s limited release of its new Claude Mythos Preview model to a select group of technology companies. Anthropic said it restricted access because of the frontier model’s ability to identify and potentially exploit software vulnerabilities. 

A potential White House executive order would aim to give AI developers a clearer process for handling future models that may create cybersecurity or other risks before public release.  

“We’re studying, possibly an executive order, to give a clear roadmap to everybody about how this is going to go and how future AIs that also potentially create vulnerabilities should go through a process so that they’re released in the wild after they’ve been proven safe, just like an FDA drug,” Hassett said during an appearance on Fox Business. 

The official said that the order – or other tools the White House may use – would likely first be applied to Mythos. He added that the issue is “pretty much what we’re working [on] almost full time … right now.” 

In the meantime, Hassett said National Cyber Director Sean Cairncross is coordinating with Anthropic to ensure that before Mythos is “released out into the wild, that it’s been tested left and right, to make sure that it doesn’t cause any harm to the American businesses or the American government.” 

An order directing increased testing for AI safety would be a shift under the Trump administration’s recent policy stance. It would also come after a highly publicized clash between the administration and Anthropic over the company’s restrictions on how the Pentagon could use its AI models, including limits related to mass surveillance and fully autonomous weapons. 

Those restrictions landed Anthropic on the federal government’s blacklist after President Donald Trump directed all agencies to cease their use of Anthropic’s AI models. But Hassett’s remarks Wednesday, alongside recent comments from Trump, suggest the administration is rethinking its relationship with Anthropic.  

Trump told CNBC’s “Squawk Box” last month that his administration has been in talks with Anthropic and that “we get along with them.”