The U.S. Department of Defense is now using artificial intelligence to draft reports it is legally required to submit to Congress, according to Ars Technica. The Pentagon is not just experimenting with the technology — it is openly boasting about it.
Alongside that disclosure, the Defense Department claims that 1.5 million of its personnel are actively using generative AI tools, according to Ars Technica. That figure, if accurate, would make the U.S. military one of the largest institutional adopters of generative AI on the planet.
The Pentagon's embrace of AI for official reporting raises immediate questions about oversight and accountability. Congressional mandates exist precisely so that lawmakers can scrutinize the military's activities with independent, human-authored assessments. If the documents Congress receives are themselves AI-generated, critics may question whether those reports reflect genuine human judgment or merely automated summarization of information the Pentagon selects to share.
There is also a transparency dimension: legislators and the public may not know which reports are AI-drafted unless the Defense Department discloses that clearly. The use of generative AI at this scale inside the military also has implications for security, accuracy, and the risk of confidently stated errors — a well-documented weakness of large language models.
This matters because the Pentagon's posture signals that AI-authored government documents are no longer hypothetical — they are already in circulation, and the rules governing them have not kept pace.