From 'Governable' to 'Any Lawful Use': The Regulatory Architecture Behind AI-Assisted Targeting

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This working paper documents the sequential weakening of U.S. civilian AI oversight constraints between January 2025 and September 2026, and maps the governance framework under which AI-assisted targeting operated at industrial scale during Operation Epic Fury (13,000 targets in 38 days). It traces the DOD language shift from 'governable' to 'any lawful use, free from usage policy constraints,' the Anthropic supply-chain risk designation and resulting judicial split (N.D. Cal vs. DC Circuit), and the European governance gap for military AI — including why IHL, GDPR, and the NATO AI Strategy do not close it. Designed as a governance companion to Airwars' AI Kill Chain (2026). Two rounds of adversarial review on a non-Anthropic model. FACT/INFERENCE methodology with differentiated source tags (FACT, PRESS REPORT, CONGRESSIONAL SOURCE, INFERENCE). Conflict of interest: drafted using Anthropic Claude; Anthropic is a subject of the analysis. Declared throughout.