Report: America's most consequential AI policy is a fully secret "voluntary" federal framework

Miles_Brundage · x · 2026-11-16

Nathan Calvin highlights that the most consequential US AI policy to date is a completely secret federal framework that loudly claims to be voluntary, while everyone involved believes it isn't.

A quoted post adds that the public underestimates how involved the federal government already is: labs must convince the feds a new model is safe before launch or face export-control consequences.

The attached screenshot shows the framework text: developers would give the government access to "covered frontier models" up to 30 days pre-release, while the text explicitly disclaims any mandatory licensing or preclearance requirement.

Original post →

More from Safety

Safety channel →