Pentagons Guidelines For Responsible AI Aren’t Always Followed, Report Shows
- The Pentagon is developing artificial intelligence weapons and projects that officials report do not consistently follow the agency's internal guidelines for responsible technology, according to a congressional watchdog...
- Defense Secretary Pete Hegseth has repeatedly dismissed warnings from international organizations and AI researchers—who caution that weapons operating with limited human control risk endangering civilians, soldiers, and accelerating...
- The push began in earnest with a January departmental memo declaring an intention to become an "AI-first" warfighting force.
The Pentagon is developing artificial intelligence weapons and projects that officials report do not consistently follow the agency’s internal guidelines for responsible technology, according to a congressional watchdog report published on October 14, 2026. A U.S. Government Accountability Office survey of 10 officials managing systems designed to automatically identify battlefield targets and process surveillance revealed compliance gaps as defense leadership pushes for rapid autonomous innovation.
Defense Secretary Pete Hegseth has repeatedly dismissed warnings from international organizations and AI researchers—who caution that weapons operating with limited human control risk endangering civilians, soldiers, and accelerating conflicts—to prioritize rapid deployment.
A Push for Autonomy
The push began in earnest with a January departmental memo declaring an intention to become an “AI-first” warfighting force. Last month, Hegseth announced a new warfare command focused on autonomous weapon development, including drones.
The GAO review examined 10 specific projects out of more than 300 AI initiatives underway across the Pentagon. When surveyed, six of the 10 officials stated that their teams had identified risks with using the projects “in an operational environment.”
Compliance Gaps and Safety Tenets
To mitigate such hazards, the Pentagon issued a 2021 memo outlining six basic tenets emphasizing oversight, verification, real-time monitoring, and allied cooperation.
Alissa Czyz, who directed the GAO review, noted that while the tenets function as signaling memos rather than binding department instructions, they were recognized as essential by project developers.
Most teams building the 10 reviewed projects had taken initial steps to follow the guidelines. However, four officials reported that their project requirements failed to align with responsible AI tenets, and three officials stated they did not know if their systems would be continuously monitored after deployment.
Staffing Shortages Hamper Oversight
Officials informed investigators that greater access to regulatory expertise would improve compliance. Yet the Department of the Responsible AI Division remains understaffed.
Authorized to hire 10 personnel in 2023, the division had filled only three positions by February, and the report indicates the Pentagon is unlikely to fill vacant positions or increase staffing.
Watchdog Recommendations and Defense Priorities
The watchdog agency issued two recommendations. While the department agreed with the proposals, a GAO representative confirmed that no changes had been implemented as of October 22, 2026. The Pentagon did not respond to inquiries regarding its timeline for adopting the recommendations.
Hegseth argued that standard committee and staffing schedules hinder national security progress.
Military AI is going to be a race for the foreseeable future where the risks to U.S. national security of moving too slowly outweigh the impacts of imperfect alignment.
That’s a very small number of folks tasked with a large responsibility of responsible AI throughout the department.
Alissa Czyz