Publication Type

Conference Proceeding Article

Version

publishedVersion

Publication Date

4-2026

Abstract

Large language models (LLMs) can exhibit concept-conditioned semantic divergence: common high-level cues (e.g., ideologies, public figures) elicit unusually uniform, stance-like responses that evade token-trigger audits. This behavior falls in a blind spot of current safety evaluations, yet carries major societal stakes, as such concept cues can steer content exposure at scale. We formalize this phenomenon and present RAVEN (Response Anomaly Vigilance), a black-box audit that flags cases where a model is simultaneously highly certain and atypical among peers by coupling semantic entropy over paraphrastic samples with cross-model disagreement. In a controlled LoRA fine-tuning study, we implant a concept-conditioned stance using a small biased corpus, demonstrating feasibility without rare token triggers. Auditing five LLM families across twelve sensitive topics (360 prompts per model) and clustering via bidirectional entailment, RAVEN surfaces recurrent, model-specific divergences in 9/12 topics. Concept-level audits complement tokenlevel defenses and provide a practical early-warning signal for release evaluation and post-deployment monitoring against propaganda-like influence.

Discipline

Artificial Intelligence and Robotics | Information Security

Research Areas

Software and Cyber-Physical Systems

Areas of Excellence

Digital transformation

Publication

Proceedings of the Fourteenth International Conference on Learning Representations (ICLR 2026), Rio de Janeiro, Brazil, April 23-27

First Page

1

Last Page

23

Publisher

ICLR

City or Country

Rio de Janeiro, Brazil

Download

Share

COinS