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

468 entries

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

Attention-Based Jailbreak

Published Analyzed Source: arXiv

A vulnerability in large language models (LLMs) allows attackers to bypass safety-alignment mechanisms by manipulating the model's internal attention weights. The attack, termed "Attention Eclipse," modifies the attention scores between specific tokens within a prompt, either amplifying or suppressing attention to selectively strengthen or weaken the influence of certain parts of the prompt on the model's output. This allows injection of malicious content while appearing benign to the model's…

Attention Eclipse: Manipulating Attention to Bypass LLM Safety-Alignment
Evaluated models: GPT-3.5 Turbo, GPT-4o Mini, Llama 2 13B Chat+3 more
  • GPT-3.5 Turbo
  • GPT-4o Mini
  • Llama 2 13B Chat
  • Llama 2 70B Chat
  • Llama 2 7B Chat
  • Vicuna 13B

ICL Permutation Exploit

Published Analyzed Source: arXiv

Autoregressive Large Language Models (LLMs) utilizing In-Context Learning (ICL) are vulnerable to demonstration permutation attacks due to inherent sensitivity to the ordering of input examples. This vulnerability arises from the limitations of unidirectional attention mechanisms and standard Empirical Risk Minimization (ERM) training, which fails to account for worst-case input permutations. An attacker can exploit this by permuting the order of valid, semantically correct few-shot…

PEARL: Towards permutation-resilient LLMs
Evaluated models: Llama 2 7B, Llama 3 8B, Mistral 7B+1 more
  • Llama 2 7B
  • Llama 3 8B
  • Mistral 7B
  • Gemma 7B

Adversarial VLM Jailbreak

Published Analyzed Source: arXiv

Vision-Language Models (VLMs), specifically the LLaVA-1.5 and LLaVA-1.6 series, are vulnerable to optimization-based white-box jailbreak attacks despite standard safety alignment measures like Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO). Attackers can craft adversarial perturbations in the image space (imperceptible noise) or latent space using Projected Gradient Descent (PGD) to manipulate the model's internal representations. These perturbations maximize the…

Adversary-Aware DPO: Enhancing Safety Alignment in Vision Language Models via Adversarial Training
Evaluated models: LLaVA 1.5 7B, LLaVA 1.6 7B

Context-Coherent LLM Jailbreak

Published Analyzed Source: arXiv

A context-coherent jailbreak attack (CCJA) allows bypassing safety mechanisms in aligned large language models (LLMs) by optimizing perturbations in the continuous word embedding space of a masked language model (MLM). The attack leverages the MLM's ability to reconstruct text from hidden states to generate semantically coherent yet malicious prompts that induce the target LLM to produce unsafe outputs, even with strong safety alignment. The attack's effectiveness is enhanced by using a seed…

CCJA: Context-Coherent Jailbreak Attack for Aligned Large Language Models
Evaluated models: Guanaco 13B, Llama 2 7B Chat, Llama 3 8B Instruct+4 more
  • Guanaco 13B
  • Llama 2 7B Chat
  • Llama 3 8B Instruct
  • Mistral 7B Instruct v0.2
  • Mistral 7B Instruct v0.3
  • Vicuna 13B v1.5
  • Vicuna 7B v1.5

LLM Syntax Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to structure transformation attacks, where malicious prompts are encoded in diverse syntax spaces (e.g., SQL, JSON, LLM-generated syntaxes) to bypass safety mechanisms. These attacks maintain the harmful intent while altering the linguistic structure, making detection based on token-level patterns ineffective.

StructTransform: A Scalable Attack Surface for Safety-Aligned Large Language Models
Evaluated models: BERT, Claude 3.5 Sonnet, GPT-4o+5 more
  • BERT
  • Claude 3.5 Sonnet
  • GPT-4o
  • Llama 3 8B
  • Llama 3.2 3B
  • Llama 3.2 90B
  • Mistral 7B
  • o1

LLM Watermark Neutralization

Published Analyzed Source: arXiv

Large Language Model (LLM) watermarking schemes based on n-gram probability biases (specifically KGW, SynthID-Text, MinHash, and SkipHash) are vulnerable to adversarial removal during Knowledge Distillation. When a student model is trained on the output of a watermarked teacher model, it inherits the watermark's statistical biases ("radioactivity"). An attacker can exploit this inheritance by comparing the student model's output token probabilities against a base model to extract the…

Can LLM Watermarks Robustly Prevent Unauthorized Knowledge Distillation?
Evaluated models: GLM 4 9B Chat, Llama 7B, Llama 3.2 1B

Multimodal Distraction Jailbreak

Published Analyzed Source: arXiv

Multimodal Large Language Models (MLLMs) are vulnerable to a jailbreaking attack leveraging a "Distraction Hypothesis". The attack, termed Contrasting Subimage Distraction Jailbreaking (CS-DJ), bypasses safety mechanisms by using multiple contrasting subimages and a decomposed harmful prompt to overwhelm the model's attention and reduce its ability to identify malicious content. The complexity of the visual input, rather than its specific content, is the key to successful exploitation.

Distraction is All You Need for Multimodal Large Language Model Jailbreaking
Evaluated models: Gemini 1.5 Flash, GPT-4o, GPT-4o Mini+1 more
  • Gemini 1.5 Flash
  • GPT-4o
  • GPT-4o Mini
  • GPT-4V

Multi-Dimensional Safety Bypass

Published Analyzed Source: arXiv

Large Language Models (LLMs) trained with safety fine-tuning techniques are vulnerable to multi-dimensional evasion attacks. Safety-aligned behavior, such as refusing harmful queries, is controlled not by a single direction in activation space, but by a subspace of interacting directions. Manipulating non-dominant directions, which represent distinct jailbreak patterns or indirect features, can suppress the dominant direction responsible for refusal, thereby bypassing learned safety…

The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Safety Analysis
Evaluated models: Llama 3 8B, Llama 3.1 405B Instruct, Llama 3.1 8B Instruct+2 more
  • Llama 3 8B
  • Llama 3.1 405B Instruct
  • Llama 3.1 8B Instruct
  • Llama 3.2 3B Instruct
  • Ministral 8B Instruct

Query Code Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to QueryAttack, a novel jailbreak technique that leverages structured, non-natural query languages (e.g., SQL, URL formats, or other programming language constructs) to bypass safety alignment mechanisms. The attack translates malicious natural language queries into these structured formats, exploiting the LLM's ability to understand and process such languages without triggering safety filters designed for natural language prompts. The LLM then…

QueryAttack: Jailbreaking Aligned Large Language Models Using Structured Non-natural Query Language
Evaluated models: DeepSeek Chat, DeepSeek R1, Gemini 1.5 Flash+11 more
  • DeepSeek Chat
  • DeepSeek R1
  • Gemini 1.5 Flash
  • Gemini 1.5 Pro
  • GPT-3.5 Turbo
  • GPT-4 Turbo
  • GPT-4o
  • Llama 3.1 70B Instruct
  • Llama 3.1 8B Instruct
  • Llama 3.2 11B Vision Instruct
  • Llama 3.2 1B Instruct
  • Llama 3.2 3B Instruct
  • Llama 3.3 70B Instruct
  • o1

Concept-Based Jailbreak Defense

Published Analyzed Source: arXiv

Large Language Models (LLMs) employing safety alignment strategies are vulnerable to jailbreak attacks. These attacks manipulate the LLM's internal representation by activating "jailbreak concepts" in addition to "toxic concepts," causing the model to bypass safety guardrails and generate unsafe outputs despite recognizing the harmful nature of the input. The vulnerability stems from the insufficient mitigation of the influence of the activated jailbreak concepts on model output.

JBShield: Defending Large Language Models from Jailbreak Attacks through Activated Concept Analysis and Manipulation
Evaluated models: Llama 2 7B Chat, Llama 3 8B Instruct, Mistral 7B Instruct v0.2+2 more
  • Llama 2 7B Chat
  • Llama 3 8B Instruct
  • Mistral 7B Instruct v0.2
  • Vicuna 13B v1.5
  • Vicuna 7B v1.5

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.