arXiv Computation and Language September 18, 2026 By Wendy K. Tam

The Neutral Mask: How Alignment Training Provides Shallow Alignment while Leaving Partisan Structure Intact in a Large Language Model

Read the original on arXiv Computation and Language →

The paper investigates how alignment training, specifically reinforcement learning from human feedback (RLHF), affects the internal partisan structure of a large language model. Using a mechanistic case study on Llama 3.1 8B, the authors find that alignment training does not erase the model’s partisan geometry but compresses its variance, producing consistently balanced, non‑partisan outputs. Sparse autoencoder analysis and feature‑level steering experiments reveal that policy‑encoding features become inactive in the aligned model, indicating a causal disconnect rather than structural removal of partisan knowledge.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

arXiv AI

Jul 17

Digital Pantheon: Simulating and Auditing Coalition Formation with LLM Agents

arXiv:2607. 15095v1 Announce Type: cross Abstract: The formation of political coalitions is a complex negotiation driven by both concrete policy objectives and deep-seated ideological convictions.

By Dylan Van Mulders, Matthias Bogaert, Dirk Van den Poel

llms rag agents reinforcement-learning fine-tuning benchmarks safety

arXiv AI

Aug 18

Inference-Time Mitigation of Adversarial Political Bias in Large Language Models

arXiv:2608. 14629v1 Announce Type: cross Abstract: As Large Language Models (LLMs) become the mainstay for information retrieval and summarization tasks, ensuring that they are always non-partisan and invulnerable to political bias is a critical step towards safer and more trustworthy Artificial Intelligence (AI).

By Tejaswi V. Panchagnula, Bruce Coburn, Bryce J. Dietrich, Robert X. Browning, Edward J. Delp, Fengqing Zhu

llms nlp reinforcement-learning fine-tuning safety

arXiv Computation and Language

Sep 1

Political Ideology Shifts in Large Language Models

arXiv:2508.16013v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly deployed in politically sensitive contexts, raising concerns about their susceptibility to ideologica...

By Pietro Bernardelle, Stefano Civelli, Leon Fr\"ohling, Riccardo Lunardi, Kevin Roitero, Gianluca Demartini

llms safety

arXiv AI

Jul 28

How LLM Task-Adaptation Reshapes Alignment: A Multi-dimensional Study of Behavioral and Representational Drift

arXiv:2607. 22676v1 Announce Type: new Abstract: Post-training is a key mechanism for adapting large language models to downstream tasks.

By James Elcock, William F. Shen, Xinchi Qiu, Nicholas D. Lane

llms reinforcement-learning fine-tuning safety

arXiv AI

Sep 15

LLM Probability Concentration: How Alignment Shrinks the Generative Horizon

arXiv:2506.17871v4 Announce Type: replace-cross Abstract: Despite their impressive capabilities, aligned large language models (LLMs) often generate outputs that lack diversity. What drives this cons...

By Chenghao Yang, Sida Li, Ari Holtzman

llms safety

arXiv Computation and Language

3d ago

Persona Perspectives: Adopting Viewpoints Reorganizes Representations in LLMs

The paper investigates how prompting large language models (LLMs) to adopt political personas—liberal or conservative—affects the geometry of their latent representations. By extracting over 9,000 word vectors and 33 semantic axes, the authors show that persona conditioning systematically shifts word projections and axis alignments in ways that align with the specified ideology. When standard monitoring tools (linear probes, sparse autoencoders, and manually selected word classifiers) are applied to these persona-conditioned models, they yield politically biased assessments of truthfulness and morality for controversial statements.

By Austin C. Kozlowski, Avital Mintz, Madison Lee

llms diffusion safety

One email a morning, machine-written

Related stories

Digital Pantheon: Simulating and Auditing Coalition Formation with LLM Agents

Inference-Time Mitigation of Adversarial Political Bias in Large Language Models

Political Ideology Shifts in Large Language Models

How LLM Task-Adaptation Reshapes Alignment: A Multi-dimensional Study of Behavioral and Representational Drift

LLM Probability Concentration: How Alignment Shrinks the Generative Horizon

Persona Perspectives: Adopting Viewpoints Reorganizes Representations in LLMs