mlsanityData to 14 Oct 2026

Bypassing Safety Guardrails in LLMs Using Humor

arXiv 2504.06577 ↗First posted 9 Apr 2025cs.CL cs.LGcited by 1

On the map: LLM safety and security → Safety alignment, refusal and guardrails

Abstract (arXiv)

In this paper, we show it is possible to bypass the safety guardrails of large language models (LLMs) through a humorous prompt including the unsafe request. In particular, our method does not edit the unsafe request and follows a fixed template -- it is simple to implement and does not need additional LLMs to craft prompts. Extensive experiments show the effectiveness of our method across different LLMs. We also show that both removing and adding more humor to our method can reduce its effectiveness -- excessive humor possibly distracts the LLM from fulfilling its unsafe request. Thus, we argue that LLM jailbreaking occurs when there is a proper balance between focus on the unsafe request and presence of humor.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. Builds on 20 of its 32 references that are in the map. First posted on arXiv 9 Apr 2025.

Builds on

20 of 32 references.

Closest ancestors · 1

Same direction, newest first

  1. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!5 Oct

Other references · 19

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2022Large Language Models are Zero-Shot Reasonerscited by 4,374
  3. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,780
  4. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  5. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 3,050
  6. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,651
  7. 2024A Survey on LLM-as-a-Judgecited by 1,423
  8. 2023Jailbreaking Black Box Large Language Models in Twenty Queriescited by 1,420
  9. 2023A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivitycited by 894
  10. 2024Great, Now Write an Article About That: The Crescendo Multi-Turn LLM Jailbreak Attackcited by 390
  11. 2024Jailbreak Attacks and Defenses Against Large Language Models: A Surveycited by 230
  12. 2024ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMscited by 216
7 more
  1. 2023Simulating Opinion Dynamics with Networks of LLM-based Agentscited by 154
  2. 2022Do Androids Laugh at Electric Sheep? Humor "Understanding" Benchmarks from The New Yorker Caption Contestcited by 98
  3. 2024Chain of Attack: a Semantic-Driven Contextual Multi-Turn attacker for LLMcited by 77
  4. 2023Fake Alignment: Are LLMs Really Aligned Well?cited by 39
  5. 2023The Wisdom of Partisan Crowds: Comparing Collective Intelligence in Humans and LLM-based Agentscited by 30
  6. 2024A Robot Walks into a Bar: Can Language Models Serve as Creativity Support Tools for Comedy? An Evaluation of LLMs' Humour Alignment with Comedianscited by 29
  7. 2024Large Language Models can Achieve Social Balancecited by 5

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025SafeVid: Toward Safety Aligned Video Large Multimodal Modelscited by 9

Similar, not cited

Close in wording, no citation either way

  1. 2025HumorReject: Decoupling LLM Safety from Refusal Prefix via A Little Humor
  2. 2026Refusal is Not Safety! Benchmarking Latent Safety Risks of LLM-Driven Content Humorization
  3. 2025QGuard:Question-based Zero-shot Guard for Multi-modal LLM Safety
  4. 2024Mitigating Exaggerated Safety in Large Language Models
  5. 2025Logic Jailbreak: Efficiently Unlocking LLM Safety Restrictions Through Formal Logical Expression
  6. 2025When Harmless Words Harm: A New Threat to LLM Safety via Conceptual Triggers
  7. 2025Playing the Fool: Jailbreaking LLMs and Multimodal LLMs with Out-of-Distribution Strategy
  8. 2024Fine-Tuning, Quantization, and LLMs: Navigating Unintended Outcomes