The process of declassifying national security records, which is hardly expeditious under the best of circumstances, will become slower as a result of the mandatory budget cuts known as sequestration.
Due to sequestration, “NARA has reduced funding dedicated to the declassification of Presidential records,” the National Archives and Records Administration (NARA) said in a report last week.
“Instead, NARA staff will prepare documents for declassification, in addition to their existing duties. This will slow declassification processes and delay other work, including FOIA responses and special access requests,” said the new report, which also identified several other adverse effects of the across-the-board cuts.
Meanwhile, because of the basic asymmetry between classification and declassification, there is no particular reason to expect a corresponding reduction in the rate at which new records are classified.
Classification is an integral part of the production of new national security information that cannot be deferred, while declassification is a distinct process that can easily be put on hold. Likewise, there is no dedicated budget for “classification” to cut in the way that NARA has cut declassification spending. And while Congress has erected barriers to declassification (such as the Kyl-Lott Amendment to prohibit automatic declassification of records without review), it has simultaneously allowed declassification requirements to go overlooked and unenforced.
Some declassification is actually mandated by law. A 1991 statute on the Foreign Relations of the United States series requires the Department of State to publish a “thorough, accurate, and reliable documentary record of major United States foreign policy decisions” no later than 30 years after the fact, necessitating the timely declassification of the underlying records. But law or no law, the government has not complied with this publication schedule.
At this inflection point, the choice is not between speed and safety but between ungoverned acceleration and a calculated momentum that allows our strategic AI advantage to be both sustained and secured.
Improved detection could strengthen deterrence, but only if accompanying hazards—automation bias, model hallucinations, exploitable software vulnerabilities, and the risk of eroding assured second‑strike capability—are well managed.
New initiative brings nine experts with federal government experience to work with the FAS and Tech & Society’s Beeck Center for Social Impact + Innovation, the Knight-Georgetown Institute, and the Institute for Technology Law & Policy Wednesday, June 11, 2025—Today Georgetown University’s Tech & Society Initiative and the Federation of American Scientists (FAS) announce two […]
A dedicated and properly resourced national entity is essential for supporting the development of safe, secure, and trustworthy AI to drive widespread adoption, by providing sustained, independent technical assessments and emergency coordination.