mlsanityData to 14 Oct 2026

Minionese: Comprehensive Benchmark and Mechanistic Study of Multilingual LLM Safety

arXiv 2607.10112 ↗First posted 11 Jul 2026cs.CR cs.AIcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks

also in 1 more
Abstract (arXiv)

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 textscMinionese, 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.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 9 of its 12 references that are in the map. First posted on arXiv 11 Jul 2026.

Builds on

9 of 12 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025The Geometry of Refusal in Large Language Models: Concept Cones and Representational Independence24 Feb
  2. 2023Universal and Transferable Adversarial Attacks on Aligned Language Models27 Jul

Other references · 7

  1. 2024HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusalcited by 1,429
  2. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  3. 2024JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Modelscited by 536
  4. 2024WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMscited by 496
  5. 2023Multilingual Jailbreak Challenges in Large Language Modelscited by 240
  6. 2025Refusal Direction is Universal Across Safety-Aligned Languagescited by 38
  7. 2024Towards Understanding the Fragility of Multilingual LLMs against Fine-Tuning Attackscited by 33

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Review
  2. 2025Do Methods to Jailbreak and Defend LLMs Generalize Across Languages?
  3. 2026IndicJR: A Judge-Free Benchmark of Jailbreak Robustness in South Asian Languages
  4. 2026IndicSafeEval: Safety Robustness of Large Language Models under Multilingual Persuasive Jailbreak Attacks
  5. 2025Evaluating LLMs Robustness in Less Resourced Languages with Proxy Models
  6. 2025Logic Jailbreak: Efficiently Unlocking LLM Safety Restrictions Through Formal Logical Expression
  7. 2025Speak Easy: Eliciting Harmful Jailbreaks from LLMs with Simple Interactions
  8. 2025SEALGuard: Safeguarding the Multilingual Conversations in Southeast Asian Languages for LLM Software Systems