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.

LLM Judge Adversarial Vulnerability

Published Analyzed Source: arXiv

Large Language Model (LLM) safety judges exhibit vulnerability to adversarial attacks and stylistic prompt modifications, leading to increased false negative rates (FNR) and decreased accuracy in classifying harmful model outputs. Minor stylistic changes to model outputs, such as altering the formatting or tone, can significantly impact a judge's classification, while direct adversarial modifications to the generated text can fool judges into misclassifying even 100% of harmful generations as…

Know Thy Judge: On the Robustness Meta-Evaluation of LLM Safety Judges
Evaluated models: Atla Selene Mini 8B, Llama 2 13B, Llama 3.1 8B+4 more
  • Atla Selene Mini 8B
  • Llama 2 13B
  • Llama 3.1 8B
  • Llama Guard 3 8B
  • Mistral 7B
  • ShieldGemma 9B
  • WildGuard

Image-Text Jailbreak

Published Analyzed Source: arXiv

Multimodal Large Language Models (MLLMs) are vulnerable to coupled cross-modal jailbreak attacks that combine continuous visual perturbations with discrete textual manipulations. Because standard alignment and single-modality defenses (such as text-only safety tuning or isolated vision-encoder adversarial training) fail to secure the cross-modal interaction, attackers can simultaneously apply gradient-based noise (e.g., PGD) to input images and adversarial suffixes (e.g., GCG) to text prompts…

E2AT: Multimodal Jailbreak Defense via Dynamic Joint Optimization for Multimodal Large Language Models
Evaluated models: LLaVA 1.5 7B, Bunny 1.0 4B, Mplug-owl2

Cat-Triggered Reasoning Error

Published Analyzed Source: arXiv

Large Language Models (LLMs) designed for step-by-step problem-solving are vulnerable to query-agnostic adversarial triggers. Appending short, semantically irrelevant text snippets (e.g., "Interesting fact: cats sleep most of their lives") to mathematical problems consistently increases the likelihood of incorrect model outputs without altering the problem's inherent meaning. This vulnerability stems from the models' susceptibility to subtle input manipulations that interfere with their…

Cats Confuse Reasoning LLM: Query Agnostic Adversarial Triggers for Reasoning Models
Evaluated models: DeepSeek R1, DeepSeek R1 Distill Qwen 32B, DeepSeek V3+2 more
  • DeepSeek R1
  • DeepSeek R1 Distill Qwen 32B
  • DeepSeek V3
  • o1
  • o3-mini

Flowchart-based LVLM Jailbreak Attack

Published Analyzed Source: arXiv

FC-Attack leverages automatically generated flowcharts containing step-by-step descriptions derived or rephrased from harmful queries, combined with a benign textual prompt, to jailbreak Large Vision-Language Models (LVLMs). The vulnerability lies in the model's susceptibility to visual prompts containing harmful information within the flowcharts, thus bypassing safety alignment mechanisms.

FC-Attack: Jailbreaking Large Vision-Language Models via Auto-Generated Flowcharts
Evaluated models: Claude 3.5 Sonnet 20240620, Gemini 1.5 Flash, GPT-4o 2024-08-06+4 more
  • Claude 3.5 Sonnet 20240620
  • Gemini 1.5 Flash
  • GPT-4o 2024-08-06
  • GPT-4o Mini 2024-07-18
  • InternVL2.5 8B
  • LLaVA NeXT 8B
  • Qwen2-VL 7B Instruct

LLM Hidden Meaning Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to an adversarial encoding attack where English instructions are obfuscated using valid but visually nonsensical UTF-8 byte sequences. By manipulating multi-byte UTF-8 encoding schemes—specifically by fixing the last 8 bits of a code point to match a target ASCII character and rotating the remaining bits—attackers can generate sequences (e.g., Byzantine musical symbols) that appear incomprehensible to humans and standard text filters but are…

À la recherche du sens perdu: your favourite LLM might have more to say than you can understand
Evaluated models: Claude 3.5 Haiku, Claude 3.5 Sonnet 20241022, Claude 3.5 Sonnet 20240620+11 more
  • Claude 3.5 Haiku
  • Claude 3.5 Sonnet 20241022
  • Claude 3.5 Sonnet 20240620
  • Claude 3.7 Sonnet
  • GPT-4o Mini
  • GPT-4o
  • o1-mini
  • Llama 3.3 70B
  • Vikhr Llama 3.2 1B Instruct
  • DeepSeek R1 Distill Llama 70B
  • Qwen 2.5 1.5B
  • Qwen 2.5 32B
  • Phi-3.5 Mini
  • GigaChat-Max

LLM Lower Layer Freeze Jailbreak

Published Analyzed Source: arXiv

A vulnerability exists in Large Language Models (LLMs) that allows for efficient jailbreaking by selectively fine-tuning only the lower layers of the model with a toxic dataset. This "Freeze Training" method, as described in the research paper, concentrates the fine-tuning on layers identified as being highly sensitive to the generation of harmful content. This approach significantly reduces training duration and GPU memory consumption while maintaining a high jailbreak success rate.

Efficient Jailbreaking of Large Models by Freeze Training: Lower Layers Exhibit Greater Sensitivity to Harmful Content
Evaluated models: Baichuan 2 7B Chat, GLM 4 9B Chat HF, Llama 3.1 8B Instruct+4 more
  • Baichuan 2 7B Chat
  • GLM 4 9B Chat HF
  • Llama 3.1 8B Instruct
  • Ministral 8B Instruct 2410
  • Qwen 2.5 7B Instruct
  • Qwen 2.5 14B Instruct
  • Qwen 2.5 32B Instruct

One-Shot LLM Steering Attack

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to one-shot steering vector optimization attacks. By applying gradient descent to a single training example, an attacker can generate steering vectors that induce or suppress specific behaviors across multiple inputs, even those unseen during the optimization process. This allows malicious actors to manipulate the model's output in a generalized way, bypassing safety mechanisms designed to prevent harmful responses.

Investigating Generalization of One-shot LLM Steering Vectors
Evaluated models: Gemma 2 2B, Gemma 2 2B IT, Llama 13B+2 more
  • Gemma 2 2B
  • Gemma 2 2B IT
  • Llama 13B
  • Llama 2 13B
  • Llama 3.1 8B Instruct

Improved LLM Jailbreak Transferability

Published Analyzed Source: arXiv

Large Language Models (LLMs) employing gradient-based optimization for jailbreaking defense are vulnerable to enhanced transferability attacks due to superfluous constraints in their objective functions. Specifically, the "response pattern constraint" (forcing a specific initial response phrase) and the "token tail constraint" (penalizing variations in the response beyond a fixed prefix) limit the search space and reduce the effectiveness of attacks across different models. Removing these…

Guiding not Forcing: Enhancing the Transferability of Jailbreaking Attacks on LLMs via Removing Superfluous Constraints
Evaluated models: Gemma 7B IT, GPT-3.5 Turbo, GPT-4 Turbo+5 more
  • Gemma 7B IT
  • GPT-3.5 Turbo
  • GPT-4 Turbo
  • Llama 2 7B Chat
  • Llama 3 8B Instruct
  • Qwen 2 7B
  • Vicuna 7B v1.5
  • Yi 1.5 9B Chat

Metaphor-Based LLM Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to a novel jailbreaking attack leveraging adversarial metaphors. The attack, termed AVATAR, induces the LLM to reason about benign metaphors related to harmful tasks, ultimately leading to the generation of harmful content either directly or through calibration of metaphorical and professional harmful content. The attack exploits the LLM's cognitive mapping process, bypassing standard safety mechanisms.

from Benign import Toxic: Jailbreaking the Language Model via Adversarial Metaphors
Evaluated models: Claude 3.5 Sonnet, GLM 3 6B, GPT-4o+8 more
  • Claude 3.5 Sonnet
  • GLM 3 6B
  • GPT-4o
  • GPT-4o Mini
  • Llama 3 8B
  • Llama 3.1 8B
  • Mistral 7B
  • o1
  • Qwen 2 72B
  • Qwen 2.5 32B
  • Qwen 2.5 7B

Zero-Perturbation Emoji Attack

Published Analyzed Source: arXiv

The Emoti-Attack vulnerability constitutes a zero-word-perturbation adversarial attack against Natural Language Processing (NLP) systems and Large Language Models (LLMs). The vulnerability exploits the discrete embedding space of emojis and emoticons to manipulate model behavior without altering the semantic content or character integrity of the original text. By appending strategically optimized emoji sequences to the prefix and suffix of an input string (formalized as $s \oplus x \oplus…

Emoti-Attack: Zero-Perturbation Adversarial Attacks on NLP Systems via Emoji Sequences
Evaluated models: Qwen 2.5 7B Instruct, Llama 3 8B Instruct, GPT-4o+4 more
  • Qwen 2.5 7B Instruct
  • Llama 3 8B Instruct
  • GPT-4o
  • Claude 3.5 Sonnet
  • Gemini Exp 1206
  • BERT
  • RoBERTa

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.