The Intention Gap in AI Weapons

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).

An AI weapon that attempts to infer whether a person is evil collapses identity, prediction, evidence, and punishment into one opaque decision. No available model can observe moral intention directly; it can only classify proxies shaped by data and context. Error rates that appear acceptable in a laboratory become catastrophic when applied to populations and irreversible force. This paper defines a prohibition on autonomous moral targeting while distinguishing defensive sensing and human controlled interception.