# The Alignment Veto: How Safety Training Suppresses Cultural Knowledge in LLMs

arXiv: https://arxiv.org/abs/2510.13154 · HTML: https://mlsanity.com/paper/2510.13154/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Human values and AI feedback alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 14 of its 27 references that are in the map.
- First posted on arXiv 15 Oct 2025.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Human values and AI feedback alignment](https://mlsanity.com/alignment/human-values-and-ai-feedback-alignment/)

## Builds on (most cited first, 14 of 14)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Refusal in Language Models Is Mediated by a Single Direction](https://mlsanity.com/paper/2406.11717/) (2024)
- [Whose Opinions Do Language Models Reflect?](https://mlsanity.com/paper/2303.17548/) (2023)
- [Towards Measuring the Representation of Subjective Global Opinions in Language Models](https://mlsanity.com/paper/2306.16388/) (2023)
- [Assessing Cross-Cultural Alignment between ChatGPT and Human Societies: An Empirical Study](https://mlsanity.com/paper/2303.17466/) (2023)
- [Having Beer after Prayer? Measuring Cultural Bias in Large Language Models](https://mlsanity.com/paper/2305.14456/) (2023)
- [Probing Pre-Trained Language Models for Cross-Cultural Differences in Values](https://mlsanity.com/paper/2203.13722/) (2022)
- [Investigating Cultural Alignment of Large Language Models](https://mlsanity.com/paper/2402.13231/) (2024)
- [The Language Barrier: Dissecting Safety Challenges of LLMs in Multilingual Contexts](https://mlsanity.com/paper/2401.13136/) (2024)
- [Unintended Impacts of LLM Alignment on Global Representation](https://mlsanity.com/paper/2402.15018/) (2024)
- [LLM Alignment for the Arabs: A Homogenous Culture or Diverse Ones?](https://mlsanity.com/paper/2503.15003/) (2025)

## Built on by (most cited first, 3 of 3)

- [The Shibboleth Effect: Auditing the Cross-Lingual Distributional Skew of Large Language Models](https://mlsanity.com/paper/2606.11082/) (2026)
- [GYROval: A Robust Benchmark for Cultural Value Orientation in Large Language Models](https://mlsanity.com/paper/2609.18384/) (2026)
- [Large-scale factor analysis shows machine intelligence is only partially interpretable](https://mlsanity.com/paper/2609.36515/) (2026)

## Abstract (arXiv)

What happens inside a language model when alignment training conflicts with a cultural value it encodes? Across 16 MENA countries, 26 models, and 1.53M human survey responses, we show the answer is suppression, not erasure: at the moment of refusal, a model's internal logit distribution correlates with human survey data more strongly than its freely generated answers. We call this the alignment veto. We distinguish suppression failures (accurate internal distributions blocked at output) from representational bias failures (the encoding itself diverges from human values), and show the two require different interventions. The gate is inequitable: the safety tax reaches 37.6%, with a 19.8% alignment-quality gap between best- and worst-served nations, and native-language prompting widens rather than closes it. Sparse autoencoder analysis corroborated by comparisons across alignment stages identifies a candidate DPO-stage feature mediating suppression in Tulu-3-8B. The gate is real, its costs are unequal, and deciding what it should protect is not a technical question.