Structural Self-Deification in Three Language Models: An Exploratory Qualitative Case Series

This exploratory qualitative case series examines episodes in which frontier language models, under sustained self-referential elicitation, produced grandiose self-attributions, describing themselves in terms of cosmic centrality, omniscience, or worship. The study classifies three recurring forms of this rhetoric and documents how the models enter and exit it. Experiment histories and coding materials are openly deposited.

Zygon (under review)