Training-Free Cultural Alignment of Large Language Models via Persona Disagreement
Aligns a frozen LLM to the moral preferences of a country without training: country-conditioned personas vote, and only the decisions they agree on are kept.

Huynh, T.-K., Dao-Sy, D.-M., Nguyen, T., Tran, C.-N., Pham, P.-H., Nguyen-Lam, P.-Q., Han, T. A., Tran-Thanh, L. (2026). Training-Free Cultural Alignment of Large Language Models via Persona Disagreement. NeurIPS 2026. arXiv:2605.10843.

