Axios reports the White House will exclude open-source models from an advanced
AI capabilities testing framework. The draft framework reviewed Tuesday would
treat "covered frontier models" as closed-source systems that are
state-of-the-art and pose national security risks, but provides no clear
definition of "state-of-the-art" or "national security risk." The White House
held a staff-level industry review; firms not invited remain unaware of the
framework's details. It is unclear which "trusted partners" will get early
access to advanced models or whether foreign governments qualify. Officials
reportedly urged companies to submit models near release rather than early
versions, and sources said less-advanced systems may be excluded.
