mlsanityData to 7 Oct 2026

Can Small Language Models Reliably Resist Jailbreak Attacks? A Comprehensive Evaluation

arXiv 2503.06519 ↗First posted 9 Mar 2025cs.CR cs.AIcited by 3

On the map: LLM safety and security → Jailbreaking

Abstract (arXiv)

Small language models (SLMs) have emerged as promising alternatives to large language models (LLMs) due to their low computational demands, enhanced privacy guarantees, and comparable performance in specific domains. Deploying SLMs on edge devices, such as smartphones and smart vehicles, has become a growing trend. However, the security implications of SLMs have not received as much attention as those of LLMs, particularly concerning the significant jailbreak threats they face. In this paper, we conduct the first systematic empirical study of SLMs' vulnerabilities to jailbreak attacks. Through systematic evaluation on 59 SLMs from 15 mainstream SLM families against 12 state-of-the-art jailbreak methods, we demonstrate that 61.0% of evaluated SLMs show an average ASR of more than 40% under jailbreak attacks and 37.3% of them have an ASR of more than 50% on direct harmful queries. Through correlation analysis, we identify that SLM vulnerabilities are closely related to training details (e.g., training dataset and method) rather than model size scaling. We further evaluate five defenses for jailbreak attacks, revealing that prompt-level defenses remain inconsistent across SLMs and attack methods, while model-level defense improves robustness against similar attacks yet generalizes poorly to multi-turn attacks such as Crescendo, highlighting the urgent need for security-by-design approaches in SLM development.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Placed in Jailbreaking, one of 12 directions in LLM safety and security. Builds on 46 of its 85 references that are in the map. First posted on arXiv 9 Mar 2025.

Builds on

46 of 85 references.

Closest ancestors · 10

Same direction, newest first

  1. 2024Hacc-Man: An Arcade Game for Jailbreaking LLMs24 May
  2. 2024Break the Breakout: Reinventing LM Defense Against Jailbreak Attacks with Self-Refinement23 Feb
  3. 2024Pruning for Protection: Increasing Jailbreak Resistance in Aligned LLMs Without Fine-Tuning19 Jan
  4. 2024How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMs12 Jan
  5. 2023DeepInception: Hypnotize Large Language Model to Be Jailbreaker6 Nov
  6. 2023Jailbreaking Black Box Large Language Models in Twenty Queries12 Oct
  7. 2023SmoothLLM: Defending Large Language Models Against Jailbreaking Attacks5 Oct
  8. 2023Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM18 Sep
  9. 2023Detecting Language Model Attacks with Perplexity27 Aug
  10. 2023GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher12 Aug

Other references · 36

  1. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  4. 2023Mamba: Linear-Time Sequence Modeling with Selective State Spacescited by 4,616
  5. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  6. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,551
  7. 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,346
  8. 2015Learning both Weights and Connections for Efficient Neural Networkscited by 1,998
  9. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,883
  10. 2023WizardLM: Empowering large pre-trained language models to follow complex instructionscited by 1,116
  11. 2024The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scalecited by 1,104
  12. 2023SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shotcited by 1,090
24 more
  1. 2023Enhancing Chat Language Models by Scaling High-quality Instructional Conversationscited by 825
  2. 2023RWKV: Reinventing RNNs for the Transformer Eracited by 814
  3. 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 709
  4. 2024MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategiescited by 694
  5. 2023Baseline Defenses for Adversarial Attacks Against Aligned Language Modelscited by 608
  6. 2024Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Researchcited by 475
  7. 2024DataComp-LM: In search of the next generation of training sets for language modelscited by 381
  8. 2022The Stack: 3 TB of permissively licensed source codecited by 354
  9. 2020The Pushshift Reddit Datasetcited by 289
  10. 2023Exploiting Programmatic Behavior of LLMs: Dual-Use Through Standard Security Attackscited by 268
  11. 2021Gradient-based Adversarial Attacks against Text Transformerscited by 233
  12. 2024Griffin: Mixing Gated Linear Recurrences with Local Attention for Efficient Language Modelscited by 232
  13. 2019BPE-Dropout: Simple and Effective Subword Regularizationcited by 189
  14. 2023CulturaX: A Cleaned, Enormous, and Multilingual Dataset for Large Language Models in 167 Languagescited by 172
  15. 2023OpenWebMath: An Open Dataset of High-Quality Mathematical Web Textcited by 151
  16. 2023RAIN: Your Language Models Can Align Themselves without Finetuningcited by 149
  17. 2024A Comprehensive Survey of Small Language Models in the Era of Large Language Models: Techniques, Enhancements, Applications, Collaboration with LLMs, and Trustworthinesscited by 125
  18. 2024Small Language Models: Survey, Measurements, and Insightscited by 109
  19. 2023LLM360: Towards Fully Transparent Open-Source LLMscited by 100
  20. 2024RecurrentGemma: Moving Past Transformers for Efficient Open Language Modelscited by 61
  21. 2024RedAgent: Red Teaming Large Language Models with Context-aware Autonomous Language Agentcited by 39
  22. 2024Hippocrates: An Open-Source Framework for Advancing Large Language Models in Healthcarecited by 15
  23. 2023MindLLM: Pre-training Lightweight Large Language Model from Scratch, Evaluations and Domain Applicationscited by 11
  24. 2024Cross-Task Defense: Instruction-Tuning LLMs for Content Safetycited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 2.

3 papers.

  1. 2026CSSBench: Evaluating the Safety of Lightweight LLMs against Chinese-Specific Adversarial Patternscited by 5
  2. 2025EASE: Practical and Efficient Safety Alignment for Small Language Modelscited by 2
  3. 2026GUARD-SLM: Token Activation-Based Defense Against Jailbreak Attacks for Small Language Modelscited by 1