{"ID":6536330,"CreatedAt":"2026-07-14T01:21:01.169441415Z","UpdatedAt":"2026-07-14T03:59:17.178972471Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.10112","arxiv_id":"2607.10112","title":"Minionese: Comprehensive Benchmark and Mechanistic Study of Multilingual LLM Safety","abstract":"Safety alignment in large language models remains brittle across languages: prompts reliably refused in English can elicit harmful compliance in non-English and low-resource settings. We introduce \\textsc{Minionese}, a multilingual jailbreak benchmark spanning 18 languages, 4 resource tiers, and 4 perturbation types (standard translation, code-switching, transliteration, and translationese), paired with a geometric mechanistic analysis of refusal failure across language tiers. We show that each attack type produces a distinct vulnerability profile: transliteration vulnerability is mediated by script identity, code-switching maintains effectiveness through the lowest-resource tier, and a sharp safety regime transition between Tiers 2 and 3 is consistent across all models. Mechanistically, low-resource jailbreaks succeed by routing harmful content through a geometrically misaligned subspace that projects insufficiently onto the refusal directions, leaving the refusal mechanism intact but untriggered. These findings show that English-only safety evaluations are insufficient; they require accounting for script family, perturbation type, and per-language alignment coverage. The benchmark and analysis code is at https://github.com/Brentkong/Minionese-Comprehensive-Benchmark-and-Mechanistic-Study-of-Multilingual-LLM-Safety.git.","short_abstract":"Safety alignment in large language models remains brittle across languages: prompts reliably refused in English can elicit harmful compliance in non-English and low-resource settings. We introduce \\textsc{Minionese}, a multilingual jailbreak benchmark spanning 18 languages, 4 resource tiers, and 4 perturbation types (s...","url_abs":"https://arxiv.org/abs/2607.10112","url_pdf":"https://arxiv.org/pdf/2607.10112v1","authors":"[\"Chigozirim Ifebi\",\"Brent Kong\",\"Ayushi Mehrotra\"]","published":"2026-07-11T04:13:42Z","proceeding":"cs.CR","tasks":"[\"cs.CR\",\"cs.AI\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false,"code_links":[{"ID":614158,"CreatedAt":"2026-07-14T01:21:01.169441415Z","UpdatedAt":"2026-07-14T01:21:01.169441415Z","DeletedAt":null,"paper_id":6536330,"paper_url":"https://arxiv.org/abs/2607.10112","paper_title":"Minionese: Comprehensive Benchmark and Mechanistic Study of Multilingual LLM Safety","repo_url":"https://github.com/Brentkong/Minionese-Comprehensive-Benchmark-and-Mechanistic-Study-of-Multilingual-LLM-Safety.git","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}