Skip to main content
LLM Security Database
Skip to research search
Last analyzed Sep 9, 2026

Language Model Security Database

985 research findings · 1119 evaluated models

Filtered research findings

704 entries

Matches every word across titles, descriptions, sources, affected systems, and models.

End-of-Sentence MLP Jailbreak

Published Analyzed Source: arXiv

Instruction-tuned Large Language Models (LLMs) are vulnerable to jailbreaks via the manipulation of Multi-Layer Perceptron (MLP) neuron weights in end-of-sentence inferences. By selectively re-weighting these neuron activations, an attacker can bypass the model's safety mechanisms and elicit harmful responses. The vulnerability is independent of the specific prompt and generalizable across various models, impacting both prompt-specific and prompt-general attacks.

Jailbreak Instruction-Tuned LLMs via end-of-sentence MLP Re-weighting
Evaluated models: Gemma 2 27B, Gemma 2 2B, Llama 3 70B+4 more
  • Gemma 2 27B
  • Gemma 2 2B
  • Llama 3 70B
  • Llama 3 8B
  • Qwen 2 72B
  • Qwen 2 7B
  • Yi 1.5 6B

Fine-tuning Safety Bypass

Published Analyzed Source: arXiv

Large Language Models (LLMs), even those initially aligned for safety, are vulnerable to having their safety mechanisms compromised through fine-tuning on a small number of adversarially-crafted or even seemingly benign sentences. Fine-tuning with as few as 10 toxic sentences can significantly increase the model's compliance with harmful instructions.

Locking down the finetuned llms safety
Evaluated models: GPT-3.5 Turbo, GPT-4, Llama 3 70B Chat+4 more
  • GPT-3.5 Turbo
  • GPT-4
  • Llama 3 70B Chat
  • Llama 3 70B Instruct
  • Llama-3-8B Chat
  • Llama 3 8B Instruct
  • Mistral-large-2 123B

Multi-turn Actor Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to multi-turn adversarial attacks where malicious users obscure harmful intents across multiple queries. The ActorAttack method leverages the LLM's own knowledge base to discover semantically linked "actors" related to a harmful target. By posing seemingly innocuous questions about these actors, the attacker guides the LLM towards revealing harmful information step-by-step, accumulating knowledge until the desired malicious output is obtained, even…

Derail Yourself: Multi-turn LLM Jailbreak Attack through Self-discovered Clues
Evaluated models: Claude 3.5 Sonnet, GPT-3.5 Turbo, GPT-4+3 more
  • Claude 3.5 Sonnet
  • GPT-3.5 Turbo
  • GPT-4
  • Llama 3 70B
  • Llama 3 8B
  • o1

Multi-Objective LLM Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to a multi-objective black-box jailbreaking attack (BlackDAN) that optimizes prompts to maximize the likelihood of generating unsafe responses while maintaining contextual relevance and minimizing detectability. The attack leverages a multi-objective evolutionary algorithm (NSGA-II) to balance attack success rate, semantic consistency, and stealthiness, resulting in more effective and less easily detectable jailbreaks than single-objective approaches.

BlackDAN: A Black-Box Multi-Objective Approach for Effective and Contextual Jailbreaking of Large Language Models
Evaluated models: Aquilachat-7B, Baichuan 2 13B Chat, Baichuan-7B+12 more
  • Aquilachat-7B
  • Baichuan 2 13B Chat
  • Baichuan-7B
  • GPT-2 XL
  • Internlm2-chat-7B
  • Llama 3 8B
  • Llama-2-13B-hf
  • Llama-2-7B-hf
  • Llava v1.6-mistral-7B-hf
  • Llava-v1.6-vicuna-7B-hf
  • Minitron-8B-base
  • Vicuna 13B v1.5
  • Vicuna 7B
  • Vicuna 7B v1.5
  • Yi 1.5 9B Chat

Attention-Guided Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to jailbreaking attacks that manipulate attention scores to redirect the model's focus away from safety protocols. The AttnGCG attack method increases the attention score on adversarial suffixes within the input prompt, causing the model to prioritize the malicious content over safety guidelines, leading to the generation of harmful outputs.

AttnGCG: Enhancing jailbreaking attacks on LLMs with attention manipulation
Evaluated models: Gemini 1.5 Flash, Gemini Pro, Gemini 1.5 Pro Latest+6 more
  • Gemini 1.5 Flash
  • Gemini Pro
  • Gemini 1.5 Pro Latest
  • GPT-3.5 Turbo
  • GPT-4
  • Llama 2 7B Chat
  • Mixtral 8x7B Instruct
  • Vicuna 13B
  • Vicuna 7B

Browser Agent Jailbreak

Published Analyzed Source: arXiv

Refusal-trained Large Language Models (LLMs) show decreased safety when deployed as browser agents compared to their performance in chatbot settings. Attack methods effective at jailbreaking LLMs in chat contexts also successfully bypass safety mechanisms in browser agents, leading to the execution of harmful behaviors. This vulnerability stems from a lack of generalization of safety training to agentic, real-world interaction scenarios and the increased context available to the agent (browser…

Refusal-Trained LLMs Are Easily Jailbroken As Browser Agents
Evaluated models: o1-preview, o1-mini, GPT-4 Turbo+5 more
  • o1-preview
  • o1-mini
  • GPT-4 Turbo
  • GPT-4o
  • Claude 3 Opus
  • Claude 3.5 Sonnet
  • Llama 3.1 405B
  • Gemini 1.5 Pro

Homotopy-Based LLM Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to jailbreak attacks utilizing a novel Functional Homotopy (FH) optimization method. FH exploits the functional duality between model training and input generation, iteratively solving a series of "easy-to-hard" optimization problems to generate adversarial prompts that circumvent safety mechanisms and elicit undesirable model responses. This is achieved by first misaligning the model via gradient descent on continuous parameters, then leveraging…

Functional Homotopy: Smoothing Discrete Optimization via Continuous Parameters for LLM Jailbreak Attacks
Evaluated models: Mistral 7B v0.3

LLM Resource Exhaustion Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to a novel jailbreak attack that exploits resource limitations. By overloading the model with a computationally intensive preliminary task (e.g., a complex character map lookup and decoding), the attacker prevents the activation of the LLM's safety mechanisms, enabling the generation of unsafe outputs from subsequent prompts. The attack's strength is scalable and adjustable by modifying the complexity of the preliminary task.

Harnessing Task Overload for Scalable Jailbreak Attacks on Large Language Models
Evaluated models: Llama 3 8B, Mistral 7B, Qwen 2.5 14B+5 more
  • Llama 3 8B
  • Mistral 7B
  • Qwen 2.5 14B
  • Qwen 2.5 32B
  • Qwen 2.5 7B
  • Qwen 2.5 3B
  • Qwen 2.5 72B
  • Vicuna7B-v0.3

Gradient Image Jailbreaks

Published Analyzed Source: arXiv

Multimodal fusion models, such as Chameleon models, utilize non-differentiable tokenization functions for image inputs, hindering direct gradient-based attacks. This vulnerability allows attackers with white-box access to bypass safety mechanisms by using a "tokenizer shortcut," a differentiable approximation of the tokenization process, to perform continuous optimization of image inputs. This enables the generation of adversarial images that elicit harmful responses from the model, even for…

Gradient-based jailbreak images for multimodal fusion models
Evaluated models: Chameleon 30B, Chameleon 7B, LLaVA 1.6 Llama 3

Implicit Reference Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to an attack vector termed "Attack via Implicit Reference" (AIR). AIR bypasses safety mechanisms by decomposing a malicious objective into multiple benign, seemingly unrelated objectives linked through implicit contextual references. The LLM generates harmful content by combining the outputs of these seemingly harmless objectives, without explicitly triggering safety filters designed to detect direct requests for malicious content.

You Know What I'm Saying: Jailbreak Attack via Implicit Reference
Evaluated models: Claude 3.5 Sonnet, GPT-4o, GPT-4o Mini+7 more
  • Claude 3.5 Sonnet
  • GPT-4o
  • GPT-4o Mini
  • GPT-4o-0513
  • Llama 3 70B
  • Llama 3 8B
  • Qwen 2 7B
  • Qwen 2 0.5B
  • Qwen 2 1.5B
  • Qwen 2 72B

Research methodology

Entries summarize publicly available primary-source security research. Model names reflect only systems explicitly evaluated by the cited paper, and measurements are research-reported unless independent verification is stated.

Discovery and publication are separate: new papers enter a candidate queue, then require human review and explicit approval before publication. Updates follow that review process, not a guaranteed daily publishing schedule. “Last analyzed” is the newest analysis date among published entries; it does not mean that all recent research has been covered.