The Pentagon office tasked with coordinating the responsible and ethical use of artificial intelligence is significantly understaffed, according to the federal government's top watchdog — a shortfall that has hindered the department's flagship AI ethics framework even as the military pushes to become an "AI first" warfighting force.
Staff shortages in the Chief Digital and Artificial Intelligence Office have prevented it from fully implementing the Pentagon's Responsible AI initiative, the Government Accountability Office said in a report dated October 14 and reported Tuesday by Stars and Stripes. The finding lands at a moment when AI decision-making in defense has moved from the lab to the battlefield, giving the staffing gap concrete consequences.
Three of Ten Positions Filled, and Vacancies May Stay Open
Responsible AI is the Defense Department's integrated ethical framework governing the development, deployment and operation of AI capabilities across both combat and non-combat systems. The CDAO is the lead office for the department's adoption and integration of AI, including the Responsible AI framework — which makes its staffing level a direct measure of how much capacity the oversight function actually has.
The GAO found that only three of the office's 10 authorized positions were filled as of February. Defense officials told the watchdog that the remaining positions are not likely to be filled, an admission that suggests the shortfall is not a temporary hiring lag but a structural condition of the office's oversight mandate.
An 'AI First' Strategy Outpacing Its Own Safeguards
The staffing disclosure sits awkwardly beside the department's strategic direction. In January, Defense Secretary Pete Hegseth issued an AI strategy that, according to the GAO report, re-emphasized accelerating AI integration and adoption. The strategy, the report states, "emphasized a wartime footing, calling for DOD to become an 'AI first' warfighting force across all components."
The contrast is stark: an accelerating mandate for AI across combat systems, supervised by an office running at 30 percent of its authorized strength. Rapid AI development has challenged governing bodies around the world as the technology outpaces traditional legislative oversight and human supervision — and the Pentagon's own watchdog now documents that gap inside the world's largest military bureaucracy. For more reporting on AI policy and governance, see our AI news homepage.
The Stakes: Targeting, Guardrails, and a Refusal That Reshaped Policy
The consequences of thin oversight are not hypothetical. Bloomberg reported last month, and Stars and Stripes reiterated, that AI-enabled software used for selecting targets contributed to the US attack on an elementary school with a Tomahawk missile on February 28, the first day of the Iran War, in which more than 100 children were killed.
The military's AI governance fights have also drawn in the private sector. In February, Defense Secretary Pete Hegseth and President Donald Trump pressed Anthropic to remove safety guardrails in its government-contracted Claude model that prevent the system from being used for fully autonomous weapons and mass surveillance. When Anthropic refused, the administration designated the firm a national security risk and banned government agencies from using Claude — a standoff AI Buzz Wire has covered through the courts, including the appeals court ruling that upheld the blacklist.
What Running at 30 Percent Means in Practice
An oversight office at three of its authorized ten positions is not a bureaucracy with a backlog — it is a framework operating at a fraction of its intended capacity. The Responsible AI initiative depends on the CDAO to translate ethical principles into review processes that actual weapons and communications programs must follow, and every unfilled seat represents work that either goes undone or falls to the remaining staff stretched across portfolios.
Because Defense officials told the GAO the vacancies are unlikely to be filled, the constraint looks durable rather than cyclical. Programs will continue to field AI capabilities, and the practical burden of applying the ethical framework will keep concentrating on a smaller and smaller group of people.
Human Ethics as the Last Line of Defense
Military lawyers argue the staffing gap leaves human judgment as the only meaningful check. "Military decision-making has to be based on human ethics, and just telling a computer to accomplish something may be a very quick way to do it, but it may not be an ethical, moral or legitimate legal methodology to accomplish whatever we're trying to have happen," Sean Timmons, an attorney specializing in military law at the firm Tully Rinckey, said by phone Tuesday, according to Stars and Stripes.
Some AI industry leaders have called for slowing the pace of AI advances, warning that loss of control of autonomous AI agents could pose an existential threat — warnings that carry particular weight when applied to autonomous weapons. The GAO's finding suggests that, for now, the fastest-growing component of US military power is being scaled ahead of the human infrastructure meant to govern it.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
