AI use by the US government is ballooning. And the lack of transparency is troubling | Nathan E Sanders and Bruce Schneier
TL;DR
On April 14, OMB disclosed 3,611 active or planned AI use cases across the US federal government, about 70% more than the final Biden-era inventory. Examples include translation tools, prison risk scoring, Palantir-backed grant screening, AI support for the veterans crisis line, and tests around autonomous nuclear reactor control. Sanders and Schneier argue the core problem is not every AI use case, but weak disclosure: many entries are one-sentence descriptions with little context or public input.
Nauti's Take
A 3,611-item inventory is a useful start; hidden automation would be worse. But one-line descriptions are not accountability when the system touches prisons, crisis calls, benefits, or reactor safety.
Government AI should face a higher bar than corporate AI: impact review, notice, appeal, and a named human owner before sensitive deployment. Otherwise efficiency becomes a polite label for moving responsibility out of sight.
Briefingshow
Government AI is not a normal product rollout. When models influence prison placement, benefits, health, safety, or crisis response, citizens need clear rights: who decides, which data matters, how errors get fixed, and who remains politically accountable. Without that, automation gains state authority without real accountability.
Sources
