Pentagon officials say AI is drafting reports for Congress
Defense officials cite major time savings from GenAI.mil, but oversight questions remain around accuracy in reports Congress uses to monitor the military.
By Maya Lindqvist · Senior Technology Correspondent
3 min read
Defense Department officials say generative AI is being used to draft reports that Congress requires from the Pentagon, a task tied directly to legislative oversight of the military. The push shows how quickly the department is applying AI to administrative work that has long strained staff time.
Emil Michael, the Pentagon’s chief technology officer, described the practice during a June 12 event hosted by the Hudson Institute in Washington, according to C-SPAN footage of the event. Michael said a congressional report that would have taken about 200 hours of staff work could be drafted in about five hours after loading relevant papers into an AI system.
The department has made generative AI tools available through GenAI.mil, its in-house platform, starting with Google Cloud’s Gemini for Government, according to a Defense Department release. The platform has been available across all six military branches since December 2025, the department said.
Short-staffed teams turn to GenAI.mil
Jacob Glassman, deputy assistant secretary of defense for science and technology foundations, gave another example at the Box Federal Summit in Washington on April 23, according to DefenseScoop. Glassman said he told a short-staffed team working on a required report to use GenAI.mil and do the best it could.
DefenseScoop reported that Glassman said the team returned a week later and described the AI-assisted product as the best report it had written in five years. Glassman did not name the report, according to DefenseScoop.
The Pentagon faces a large and growing reporting load. The Government Accountability Office said the number of recurring Defense Department reports to Congress rose from just over 500 in 2000 to more than 1,400 by 2020.
Elizabeth Field, a former GAO senior executive director, told Federal News Network in 2023 that Pentagon legislative affairs officials often had to review defense authorization statutes nearly line by line to identify new reporting duties. A GAO report found that identifying those duties and assigning them to offices could take three to six months, even though some reports were due within a year.
Accuracy questions follow wider AI rollout
The officials’ remarks focused on speed, but they did not describe a process for checking the accuracy of AI-drafted reports before they reach Congress. Such reports help lawmakers assess defense programs, spending and military policy.
Other organizations have run into problems after relying on AI-generated text without enough review. The Financial Times reported that KPMG withdrew a report on AI in business after the research group GPTZero found case studies containing AI-generated errors and false claims.
Generative AI use inside the military is not limited to congressional reports. A Small Wars Journal article said service members have used such tools to draft personnel evaluations for officers and non-commissioned officers, write commendation medal citations and produce counseling statements.
Michael said at the Hudson Institute event that Defense Department use of commercial AI tools through GenAI.mil rose from 80,000 personnel in December 2025 to 1.5 million in June 2026. The department’s workforce is about 3.5 million people.
Google was among several technology companies that signed 2025 agreements with the General Services Administration to offer AI tools across federal agencies at discounted prices, according to GSA announcements. On May 1, the Defense Department also announced agreements with SpaceX, OpenAI, Google, Nvidia, Reflection AI, Microsoft, Amazon Web Services and Oracle to put AI tools on classified networks for lawful operational use.
The government has not disclosed how much it is paying under those contracts. The Defense Department list did not include Anthropic, which Ars Technica reported had been blacklisted by the Trump administration after the company allegedly refused unrestricted use of its Claude models for autonomous warfare and mass surveillance.
This story draws on original reporting from Ars Technica.