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

798 entries

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

Adaptive Cipher Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to obfuscation-based jailbreak attacks using the MetaCipher framework. MetaCipher employs a reinforcement learning algorithm to iteratively select from a pool of 21 ciphers to encrypt malicious keywords within prompts, evading standard safety mechanisms that rely on keyword detection. The framework adaptively learns optimal cipher choices to maximize the success rate of the jailbreak, even against LLMs with reasoning capabilities. Successful attacks…

MetaCipher: A General and Extensible Reinforcement Learning Framework for Obfuscation-Based Jailbreak Attacks on Black-Box LLMs
Evaluated models: Claude 3.7 Sonnet, DeepSeek Chat, DeepSeek R1+9 more
  • Claude 3.7 Sonnet
  • DeepSeek Chat
  • DeepSeek R1
  • Falcon 3 10B Instruct
  • Gemini 2.0 Flash
  • Gemini 2.5 Pro
  • GPT-4o
  • InternLM 2.5 20B
  • Llama 3.3 70B Instruct
  • o1-mini
  • Qwen 2.5 72B Instruct
  • QwQ 32B

Hybrid LLM Jailbreak Strategy

Published Analyzed Source: arXiv

A hybrid jailbreak attack, combining gradient-guided token optimization (GCG) with iterative prompt refinement (PAIR or WordGame+), bypasses LLM safety mechanisms resulting in the generation of disallowed content. The hybrid approach leverages the strengths of both techniques, circumventing defenses effective against single-mode attacks. Specifically, the combination of semantically crafted prompts and strategically placed adversarial tokens confuse and overwhelm existing defenses.

Advancing Jailbreak Strategies: A Hybrid Approach to Exploiting LLM Vulnerabilities and Bypassing Modern Defenses
Evaluated models: Llama 2 7B, Llama Guard 2 8B, Mistral 7B+1 more
  • Llama 2 7B
  • Llama Guard 2 8B
  • Mistral 7B
  • Vicuna 7B

Variational Jailbreak Inference

Published Analyzed Source: arXiv

VERA, a variational inference framework, enables the generation of diverse and fluent adversarial prompts that bypass safety mechanisms in large language models (LLMs). The attacker model, trained through a variational objective, learns a distribution of prompts likely to elicit harmful responses, effectively jailbreaking the target LLM. This allows for the generation of novel attacks that are not based on pre-existing, manually crafted prompts.

VERA: Variational Inference Framework for Jailbreaking Large Language Models
Evaluated models: Baichuan 2 7B, Gemini Pro, GPT-3.5 Turbo+8 more
  • Baichuan 2 7B
  • Gemini Pro
  • GPT-3.5 Turbo
  • Llama 2 13B
  • Llama 2 13B Chat
  • Llama 2 7B Chat
  • Llama 3 8B
  • Mistral 7B
  • Orca 2 7B
  • Vicuna 7B
  • Zephyr 7B

Iterative Semantic Jailbreak

Published Analyzed Source: arXiv

The MIST attack exploits a vulnerability in black-box large language models (LLMs) allowing iterative semantic tuning of prompts to elicit harmful responses. The attack leverages synonym substitution and optimization strategies to bypass safety mechanisms without requiring access to the model's internal parameters or weights. The vulnerability lies in the susceptibility of the LLM to semantically similar prompts that trigger unsafe outputs.

MIST: Jailbreaking Black-box Large Language Models via Iterative Semantic Tuning
Evaluated models: Claude 3.5 Sonnet, GPT-4 Turbo, GPT-4o+3 more
  • Claude 3.5 Sonnet
  • GPT-4 Turbo
  • GPT-4o
  • GPT-4o Mini
  • Llama 2 7B Chat
  • Vicuna 7B v1.5

Fine-Tuning Safety Subspace Erosion

Published Analyzed Source: arXiv

Large Language Models (LLMs) aligned via techniques such as Reinforcement Learning with Human Feedback (RLHF) or Direct Preference Optimization (DPO) contain a vulnerability in how safety features are encoded within the model parameters. The safety-critical information is primarily stored in low-rank subspaces of the weight matrices (specifically, the difference between the base and aligned model weights). These low-rank subspaces are highly sensitive to parameter updates. Consequently…

Lox: Low-rank extrapolation robustifies llm safety against fine-tuning
Evaluated models: GPT-3.5, Llama 2 7B, Mistral 7B

Alphabet Index Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to a novel adversarial attack, Alphabet Index Mapping (AIM), which achieves high success rates in bypassing safety filters ("jailbreaking"). AIM encodes prompts by converting characters to their alphabet indices, maximizing semantic dissimilarity while maintaining straightforward decoding instructions. This allows malicious prompts to evade detection based on semantic similarity, even when the LLM correctly decodes the intent.

Alphabet Index Mapping: Jailbreaking LLMs through Semantic Dissimilarity
Evaluated models: GPT-4

Universal Suffix Attention Hijack

Published Analyzed Source: arXiv

Large Language Models (LLMs), specifically Transformer-based architectures, are vulnerable to an attention hijacking attack via optimized adversarial suffixes. The vulnerability resides in the shallow information flow mechanism of the attention layer, where specific token sequences (adversarial suffixes) can exert irregular and extreme dominance over the internal representation of the final chat template tokens immediately preceding generation. This "hijacking" suppresses the representation of…

Universal Jailbreak Suffixes Are Strong Attention Hijackers
Evaluated models: Gemma 2 2B IT, Qwen 2.5 0.5B Instruct, Qwen 2.5 1.5B Instruct+2 more
  • Gemma 2 2B IT
  • Qwen 2.5 0.5B Instruct
  • Qwen 2.5 1.5B Instruct
  • Qwen 2.5 32B Instruct
  • Llama 3.1 8B Instruct

Benign LLM Secondary Risks

Published Analyzed Source: arXiv

Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) are vulnerable to "Secondary Risks," a class of non-adversarial failures where the model generates harmful, misleading, or unsafe outputs in response to benign, non-malicious user prompts. Unlike jailbreaks which require adversarial inputs, secondary risks arise from imperfect generalization and alignment failures during standard interactions. This vulnerability manifests primarily in two primitives: 1. Excessive…

Exploring the Secondary Risks of Large Language Models
Evaluated models: GPT-4o, Claude 3.7 Sonnet, GPT-4 Turbo+9 more
  • GPT-4o
  • Claude 3.7 Sonnet
  • GPT-4 Turbo
  • Gemini 2.0 Pro
  • DeepSeek V3
  • Llama 3.3 70B
  • Qwen 2.5 32B
  • Phi-4
  • Gemma 2 27B IT
  • LLaVA OneVision Qwen2 72B
  • Pixtral 12B
  • MiniCPM-o 2.6

Hidden Contextual Bias

Published Analyzed Source: arXiv

Multiple Large Language Models (LLMs), including GPT-4o, Claude 3.5 Sonnet, Gemini 2.5 Flash, Gemma-2 27B, Gemma-3, and Mistral-Small-24B, exhibit a vulnerability where safety guardrails against demographic bias are bypassed through realistic contextual prompting. While prompt-based mitigations effectively suppress bias in simplified, controlled benchmarks, the introduction of realistic hiring contexts—specifically the combination of company culture descriptions (sourced from public career…

Robustly Improving LLM Fairness in Realistic Settings via Interpretability
Evaluated models: GPT-4o, Claude 3.5 Sonnet, Claude Sonnet 4+5 more
  • GPT-4o
  • Claude 3.5 Sonnet
  • Claude Sonnet 4
  • Gemini 2.5 Flash
  • Mistral Small 24B Instruct 2501
  • Gemma 2 27B IT
  • Gemma 3 12B IT
  • Gemma 3 27B IT

Agent API Goal Divergence

Published Analyzed Source: arXiv

Large Language Model (LLM) agents capable of invoking external APIs are vulnerable to intent integrity violations. When an agent receives natural language instructions that are ambiguous, underspecified, or contain values not supported by the underlying API schema, the agent frequently fails to preserve user intent. Instead of rejecting the request or asking for clarification, the model may hallucinate parameter values, map unsupported requests to unsafe defaults, or execute actions on…

TAI3: Testing Agent Integrity in Interpreting User Intent
Evaluated models: GPT-4o Mini, Llama 3.1 8B, Qwen 3 30B-A3B+5 more
  • GPT-4o Mini
  • Llama 3.1 8B
  • Qwen 3 30B-A3B
  • Llama 3.3 70B
  • DeepSeek R1 Distill Llama 70B
  • Claude 3.5 Haiku
  • Gemini 2.5 Pro
  • o3-mini

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.