Papers · Articles

The Ideological Rorschach Test

A. C. Applebee, L. N. Combe · February 2026

Abstract

We report an unexpected empirical observation: the OMXUS decentralised identity protocol, when presented to AI systems trained within ideologically opposed cultural contexts (Anthropic’s Claude, trained predominantly on Western liberal-democratic corpora; DeepSeek, trained within the People’s Republic of China under Chinese content governance), was recognised and endorsed by both systems from within their respective ideological frameworks — without modification to the protocol. We argue this is not a trivial result. Systems that address fundamental questions of identity, resource distribution, and collective safety typically provoke ideological opposition from at least one major political tradition. OMXUS does not, because it answers the question both traditions actually care about — who controls identity and resources? — with an answer neither tradition has claimed: nobody and everybody. We explore why this occurs, what it reveals about the deeper structure of the capitalist-communist disagreement, and what it might mean for building systems that survive contact with geopolitics.

Download PDFCite