Michelle Beyers (Aequitas Institute; Arizona State University (ASU)) has posted Manufactured Legitimacy: A Structural Diagnosis of the Accountability Gap in Defense Artificial Intelligence Governance on SSRN. Here is the abstract:
The United States Department of Defense operationalizes its mandatory ethical principles through the National Institute of Standards and Technology’s voluntary Artificial Intelligence Risk Management Framework, its primary implementation tool. This article argues the instrument cannot govern military artificial intelligence deployed under the structural conditions that define consequential military use: operational security, classification, command and control, weapons integration, and mission-critical operation. Applying a 3+1 structural accountability framework developed through analysis of post-conflict policing governance, the article identifies three conditions (a visibility boundary, institutional self-definition, and an evidentiary constraint) that produce accountability gaps that rigorous compliance cannot close, plus a legitimization function that converts the gap into institutional legitimacy. The gap follows from the environment, not the instrument’s design; the Department adopted the instrument before it was finalized, without a documented fitness assessment. The article proposes a precondition model of legitimacy that displaces the dominant output model.
Recommended!