Pentagon says AI use surged, though adoption is still under half

Pentagon Chief Technology Officer Emil Michael said Defense Department use of commercial AI tools has climbed to 1.5 million personnel, a rapid increase that officials cite as evidence of faster government adoption. Using the department’s workforce figure of about 3.5 million people, the figure still puts AI use at roughly 43%, leaving most personnel outside the reported user base.

Michael gave the update at an event hosted by the Hudson Institute, where he said the number of Pentagon personnel using commercial AI tools had risen from 80,000 in December 2025 to 1.5 million this month. That change amounts to an increase of more than 18-fold over the period.

The growth follows a broader federal push to use artificial intelligence for administrative work, including goals associated with the Department of Government Efficiency. The Office of Management and Budget disclosed more than 3,600 active or planned federal AI use cases last year, up 70% from the prior year.

Officials point to paperwork savings

Michael said one practical use is drafting mandatory reports to Congress. He told the Hudson Institute audience that AI could turn a task requiring 200 hours of staff time into a five-hour job by loading relevant documents into the system and having it produce a draft.

The reporting burden has grown over time. The U.S. Government Accountability Office said the number of Defense Department reports mandated by Congress rose from 500 in 2000 to about 1,400 in 2020.

Other Pentagon officials have made similar claims about time savings. DefenseScoop reported that Jacob Glassman, the deputy assistant secretary of defense for science and technology foundations, said at a Box Federal Summit that a short-staffed team used GenAI.mil, the Pentagon’s enterprise AI tool, to create a congressional report in about a week; Glassman said the team described it as its best report in five years.

The Defense Department has also expanded access to outside AI products. In May, the Pentagon announced agreements involving SpaceX, OpenAI, Google, Nvidia, Reflection, Microsoft, Amazon Web Services and Oracle for operational use of their AI tools. In 2025, the General Services Administration said Google and other technology companies had signed agreements to offer tools to federal agencies at discounted prices.

Adoption gaps and risk warnings remain

Federal AI use remains uneven, according to Brookings Institution research based on jobs data, OMB memos and interviews with technologists across eight agencies through the end of 2025. Brookings found that five large agencies accounted for more than half of reported federal AI inventory use over the prior three years, while 11 small agencies reported 2% of the 2025 total.

Brookings attributed slower adoption to staffing limits, risk-averse cultures, procurement and funding problems, and low public trust in AI systems. It also said agencies often lack enough training resources to help workers test and use the tools.

Private-sector adoption shows similar limits. Goldman Sachs’ March 2026 AI Adoption Tracker said AI use was growing but remained short of becoming a standard workplace tool, while Census Bureau survey data put economywide adoption at 19.5% in May 2026 and projected 22.7% over the next six months. The Census Bureau said establishments with more than 250 employees led adoption at 36.1%.

Worker trust is another barrier. WalkMe’s April State of Digital Adoption report found that more than half of surveyed workers bypassed AI tools and did tasks manually, and 9% said they trusted AI for complex work.

The GAO has warned that AI can help agencies cross-reference data and share information, but can also produce false information intentionally or unintentionally. In a March report, the watchdog urged OMB to identify privacy risks tied to expanded AI use and give agencies guidance before sensitive information is exposed.

This story draws on original reporting from Fortune.