Overview
- Kevin Hassett likened possible AI reviews to FDA drug testing on Wednesday, and senior aides soon signaled the administration is not set on a blanket approval system.
- A White House official called reports of executive orders speculation and said any decision would come directly from the president, with drafts still under active debate.
- Options under discussion include voluntary safety testing through the Commerce Department’s Center for AI Standards and Innovation, which reviews models that companies share for evaluation.
- Some officials are weighing pre-release checks by the intelligence community so the government can study powerful tools before adversaries do, according to people familiar with the talks.
- Security worries about Anthropic’s unreleased Mythos model, which the company says can rapidly find software flaws, and an ongoing Pentagon-Anthropic clash over usage limits have sharpened the push for clearer rules.