GAO report finds Pentagon often sidesteps its own responsible AI rules

A October GAO report examined ten Pentagon AI projects, revealing that several do not consistently adhere to the six responsible-AI principles issued in 2021. Six of the ten officials surveyed reported identified risks, and four said their work was not aligned with the guidelines. The Department of the Responsible AI Division, tasked with training and oversight, has filled only three of ten authorized positions, limiting its ability to enforce compliance.

While most projects have taken preliminary steps toward the tenets, three respondents were unsure whether continuous monitoring would occur after deployment. The GAO recommended clearer responsibility assignments and staffing boosts; the Pentagon agreed but had not implemented changes as of late October. Defense Secretary Pete Hegseth has emphasized rapid autonomous innovation, arguing that slow processes threaten national security.