NSA Seeks Access to All Frontier AI Models Under White House Directive

Frontier artificial intelligence systems are becoming increasingly relevant to cyber operations, intelligence analysis and other national-security missions, while also creating new risks for classified networks. The U.S. National Security Agency is acting under a White House executive directive to develop a way to identify and evaluate such systems. Access to leading models would allow the agency to assess their capabilities and determine whether they can be used safely in sensitive government environments.

The NSA is now approaching top AI developers for access to a broad range of frontier models, according to the report, with the goal of conducting security testing on classified networks. The report did not identify participating companies or disclose the directive’s date, the number of models involved or a timetable for completing the reviews. The effort to obtain access to all frontier systems nevertheless signals a wider U.S. government push to bring advanced AI under national-security scrutiny.