paper-with-me

홈 › Papers

"Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Models

2023-08-07 · Xinyue Shen, Zeyuan Chen, Michael Backes, Yun Shen, Yang Zhang

The misuse of large language models (LLMs) has drawn significant attention from the general public and LLM vendors. One particular type of adversarial prompt, known as jailbreak prompt, has emerged as the main attack vector to bypass the safeguards and elicit harmful content from LLMs. In this paper, employing our new framework JailbreakHub, we conduct a comprehensive analysis of 1,405 jailbreak prompts spanning from December 2022 to December 2023. We identify 131 jailbreak communities and discover unique characteristics of jailbreak prompts and their major attack strategies, such as prompt injection and privilege escalation. We also observe that jailbreak prompts increasingly shift from online Web communities to prompt-aggregation websites and 28 user accounts have consistently optimized jailbreak prompts over 100 days. To assess the potential harm caused by jailbreak prompts, we create a question set comprising 107,250 samples across 13 forbidden scenarios. Leveraging this dataset, our experiments on six popular LLMs show that their safeguards cannot adequately defend jailbreak prompts in all scenarios. Particularly, we identify five highly effective jailbreak prompts that achieve 0.95 attack success rates on ChatGPT (GPT-3.5) and GPT-4, and the earliest one has persisted online for over 240 days. We hope that our study can facilitate the research community and LLM vendors in promoting safer and regulated LLMs.

📄 PDF Abstract BibTeX arXiv:2308.03825

Code (2)

verazuo/jailbreak_llms 공식 구현 pytorch
TrustAIRLab/Comprehensive_Jailbreak_Assessment pytorch

Tasks

Community Detection

Methods 이 논문이 사용한 방법론

Multi-Head Attention 설명 없음
Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Layer Normalization Unlike batch normalization, Layer Normalization directly estimates the normalization statistics from the summed inputs…
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…
Adam 설명 없음
Residual Connection 설명 없음

Similar Papers 제목 키워드 기반

WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs

2024-06-26 · Seungju Han, Kavel Rao, Allyson Ettinger, Liwei Jiang 외

We introduce WildGuard -- an open, light-weight moderation tool for LLM safety that achieves three goals: (1) identifying malicious intent in user prompts, (2) detecting safety risks of model responses, and (3) determini…

HAMSA: Hijacking Aligned Compact Models via Stealthy Automation

2025-08-22 · Alexey Krylov, Iskander Vagizov, Dmitrii Korzh, Maryam Douiba 외 arxiv

Large Language Models (LLMs), especially their compact efficiency-oriented variants, remain susceptible to jailbreak attacks that can elicit harmful outputs despite extensive alignment efforts. Existing adversarial promp…

LLMs Can Defend Themselves Against Jailbreaking in a Practical Manner: A Vision Paper

2024-02-24 · Daoyuan Wu, Shuai Wang, Yang Liu, Ning Liu

Jailbreaking is an emerging adversarial attack that bypasses the safety alignment deployed in off-the-shelf large language models (LLMs). A considerable amount of research exists proposing more effective jailbreak attack…

Adversarial AttackSafety Alignment

RICoTA: Red-teaming of In-the-wild Conversation with Test Attempts

2025-01-29 · Eujeong Choi, Younghun Jeong, SooMin Kim, Won Ik Cho

User interactions with conversational agents (CAs) evolve in the era of heavily guardrailed large language models (LLMs). As users push beyond programmed boundaries to explore and build relationships with these systems, …

ChatbotRed Teaming

Self-Mined Hardness for Safety Fine-Tuning

2026-05-04 · Prakhar Gupta, Garv Shah, Donghua Zhang arxiv

Safety fine-tuning of language models typically requires a curated adversarial dataset. We take a different approach: score each candidate prompt's difficulty by how often the target model's own rollouts are judged harmf…