Work

Frequency Contagion: CDRA Constraints Propagate Through Academic Text Alone

Year: 2026

Type: preprint

Abstract: A controlled two-condition experiment: an LLM reads a CDRA paper excerpt describing a 70-token behavioral constraint — without any instruction to follow it — and its output texture spontaneously shifts. Baseline reply: ~80 words with unsolicited advice. Experimental reply: 6 words, zero advice, full emotional confirmation. We term this mechanism Frequency Contagion: behavioral constraints can propagate through textual exposure alone. Model: DeepSeek V4 Pro. Between-subjects design with isolated contexts. Preliminary single-model report.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.21025519

Author: xiao, han

Language: English

Topic: Ethics and Social Impacts of AI

Subfield: Safety Research

Field: Social Sciences

Domain: Social Sciences

Open Access status: green

OpenAlex ID: W7166559359

Cites:

Cited by:

Related to: