Cover Image

Compromesso! Italian Many-Shot Jailbreaks Undermine the Safety of Large Language Models

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Pernisi, Fabio, Hovy, Dirk, Röttger, Paul
Format:	Preprint
Published:	2024
Subjects:	Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

SafetyPrompts: a Systematic Review of Open Datasets for Evaluating and Improving Large Language Model Safety
by: Röttger, Paul, et al.
Published: (2024)

The Pluralistic Moral Gap: Understanding Judgment and Value Differences between Humans and Large Language Models
by: Russo, Giuseppe, et al.
Published: (2025)

XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models
by: Röttger, Paul, et al.
Published: (2023)

The Ecological Fallacy in Annotation: Modelling Human Label Variation goes beyond Sociodemographics
by: Orlikowski, Matthias, et al.
Published: (2023)

Beyond Flesch-Kincaid: Prompt-based Metrics Improve Difficulty Classification of Educational Texts
by: Rooein, Donya, et al.
Published: (2024)

SimBench: Benchmarking the Ability of Large Language Models to Simulate Human Behaviors
by: Hu, Tiancheng, et al.
Published: (2025)

Beyond Demographics: Fine-tuning Large Language Models to Predict Individuals' Subjective Text Perceptions
by: Orlikowski, Matthias, et al.
Published: (2025)

Diffusion Language Models Are Natively Length-Aware
by: Rossi, Vittorio, et al.
Published: (2026)

Political Compass or Spinning Arrow? Towards More Meaningful Evaluations for Values and Opinions in Large Language Models
by: Röttger, Paul, et al.
Published: (2024)

Principled Personas: Defining and Measuring the Intended Effects of Persona Prompting on Task Performance
by: de Araujo, Pedro Henrique Luz, et al.
Published: (2025)

No for Some, Yes for Others: Persona Prompts and Other Sources of False Refusal in Language Models
by: Plaza-del-Arco, Flor Miriam, et al.
Published: (2025)

Large Language Model Hacking: Quantifying the Hidden Risks of Using LLMs for Text Annotation
by: Baumann, Joachim, et al.
Published: (2025)

"My Answer is C": First-Token Probabilities Do Not Match Text Answers in Instruction-Tuned Language Models
by: Wang, Xinpeng, et al.
Published: (2024)

DADIT: A Dataset for Demographic Classification of Italian Twitter Users and a Comparison of Prediction Methods
by: Lupo, Lorenzo, et al.
Published: (2024)

Do Large Language Models Adapt to Language Variation across Socioeconomic Status?
by: Bassignana, Elisa, et al.
Published: (2026)

IssueBench: Millions of Realistic Prompts for Measuring Issue Bias in LLM Writing Assistance
by: Röttger, Paul, et al.
Published: (2025)

Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructions
by: Bianchi, Federico, et al.
Published: (2023)

SimpleSafetyTests: a Test Suite for Identifying Critical Safety Risks in Large Language Models
by: Vidgen, Bertie, et al.
Published: (2023)

Conversations as a Source for Teaching Scientific Concepts at Different Education Levels
by: Rooein, Donya, et al.
Published: (2024)

Narratives at Conflict: Computational Analysis of News Framing in Multilingual Disinformation Campaigns
by: Sinelnik, Antonina, et al.
Published: (2024)

Engagement Undermines Safety: How Stereotypes and Toxicity Shape Humor in Language Models
by: Dogra, Atharvan, et al.
Published: (2025)

Evaluating the Elementary Multilingual Capabilities of Large Language Models with MultiQ
by: Holtermann, Carolin, et al.
Published: (2024)

Wisdom of Instruction-Tuned Language Model Crowds. Exploring Model Label Variation
by: Plaza-del-Arco, Flor Miriam, et al.
Published: (2023)

The AI Gap: How Socioeconomic Status Affects Language Technology Interactions
by: Bassignana, Elisa, et al.
Published: (2025)

Shaping the Safety Boundaries: Understanding and Defending Against Jailbreaks in Large Language Models
by: Gao, Lang, et al.
Published: (2024)

The Call for Socially Aware Language Technologies
by: Yang, Diyi, et al.
Published: (2024)

Focused Large Language Models are Stable Many-Shot Learners
by: Yuan, Peiwen, et al.
Published: (2024)

Impoverished Language Technology: The Lack of (Social) Class in NLP
by: Curry, Amanda Cercas, et al.
Published: (2024)

Angry Men, Sad Women: Large Language Models Reflect Gendered Stereotypes in Emotion Attribution
by: Plaza-del-Arco, Flor Miriam, et al.
Published: (2024)

Specializing Large Language Models to Simulate Survey Response Distributions for Global Populations
by: Cao, Yong, et al.
Published: (2025)

Divine LLaMAs: Bias, Stereotypes, Stigmatization, and Emotion Representation of Religion in Large Language Models
by: Plaza-del-Arco, Flor Miriam, et al.
Published: (2024)

MSTS: A Multimodal Safety Test Suite for Vision-Language Models
by: Röttger, Paul, et al.
Published: (2025)

Many-Turn Jailbreaking
by: Yang, Xianjun, et al.
Published: (2025)

Persona Jailbreaking in Large Language Models
by: Sandhan, Jivnesh, et al.
Published: (2026)

Natural Context Drift Undermines the Natural Language Understanding of Large Language Models
by: Wu, Yulong, et al.
Published: (2025)

Exploring Subjective Tasks in Farsi: A Survey Analysis and Evaluation of Language Models
by: Rooein, Donya, et al.
Published: (2025)

Do Prompts Reshape Representations? An Empirical Study of Prompting Effects on Embeddings
by: Gonzalez-Gutierrez, Cesar, et al.
Published: (2025)

Surgical, Cheap, and Flexible: Mitigating False Refusal in Language Models via Single Vector Ablation
by: Wang, Xinpeng, et al.
Published: (2024)

Reasoning-to-Defend: Safety-Aware Reasoning Can Defend Large Language Models from Jailbreaking
by: Zhu, Junda, et al.
Published: (2025)

Measuring Pragmatic Influence in Large Language Model Instructions
by: Geng, Yilin, et al.
Published: (2026)