Metraipolis › archive › dc_ae2cb22cb4a0
Quantifying and Mitigating Korean Jamo-Level Typographical Vulnerabilities in Large Language Models — arXiv:2608.30229
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.30229
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T08:07:25.528Z
- SHA-256 of the archived text
115ec0d91224d51b1c505515d2380a5e2a97becab2c268b23cb74257e08ba337- Stable URL of this page
https://metraipolis.com/documento/dc_ae2cb22cb4a0
How to cite
Quantifying and Mitigating Korean Jamo-Level Typographical Vulnerabilities in Large Language Models — arXiv:2608.30229 — arXiv (en), https://arxiv.org/abs/2608.30229 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ae2cb22cb4a0 (Metraipolis archive dc_ae2cb22cb4a0, sha256 115ec0d91224d51b1c505515d2380a5e2a97becab2c268b23cb74257e08ba337)
BibTeX
@misc{dcae2cb22cb4a0,
title = {Quantifying and Mitigating Korean Jamo-Level Typographical Vulnerabilities in Large Language Models — arXiv:2608.30229},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.30229},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ae2cb22cb4a0; sha256 115ec0d91224d51b1c505515d2380a5e2a97becab2c268b23cb74257e08ba337},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30229. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Korean introduces an additional typographical perturbation level not captured by ordinary character-level edit models: because syllable blocks are internally composed of sub-character units called jamo, keyboard-level errors can occur within a syllable, either producing a valid but semantically altered character or exposing raw jamo on the surface. Both outcomes disrupt sub-word tokenization and are not reliably corrected by existing grammatical error correction pipelines, leaving LLMs directly exposed to corrupted inputs. To quantify this vulnerability, we apply five jamo-level perturbation types to the KMMLU benchmark and evaluate four language models, finding that accuracy declines monotonically with perturbation intensity and that parameter scaling does not confer robustness against intra-syllabic noise. We further show that typo-corrupted inputs induce a distinct shift in internal representations that is not reducible to ordinary answer incorrectness, and that a simple linear probe trained on these representations detects unseen perturbation types with high AUROC. Motivated by this signal, we propose Typo-Aware Chain-of-Thought (TACoT), which routes inputs to chain-of-thought inference only when the probe detects a likely typo, recovering a substantial portion of the CoT accuracy gain at a fraction of the inference cost. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2608.30229 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T07:47:52.788Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).