White House Limits National Security AI Reviews to Closed-Source Frontier Models
The administration's artificial intelligence safety framework will focus on closed-source advanced models rather than open-source systems, prompting debate over how emerging security risks from open-weight technologies should be monitored.