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.

SAE Interpretability Illusion

Published Analyzed Source: arXiv

Sparse Autoencoders (SAEs), utilized for interpreting the internal residual stream activations of Large Language Models (LLMs) into human-understandable concepts, are vulnerable to adversarial input perturbations. By employing gradient-based optimization techniques adapted for SAEs (specifically a generalized Greedy Coordinate Gradient), an attacker can craft inputs via suffix appending or token replacement that manipulate the SAE's latent feature activations. This vulnerability allows for the…

Interpretability Illusions with Sparse Autoencoders: Evaluating Robustness of Concept Representations
Evaluated models: Llama 3 8B, Gemma 2 9B

Semantic Audio Jailbreak

Published Analyzed Source: arXiv

Large Audio-Language Models (LAMs) are vulnerable to adversarial signal-level perturbations that allow for the bypass of safety guardrails (jailbreaking). While these models may possess robust text-based safety alignment, they fail to generalize this robustness to the audio modality. Attackers can utilize the Audio Perturbation Toolkit (APT) to apply transformations in the time domain (Energy Distribution Perturbation, Trimming, Fade In/Out), frequency domain (Pitch Shifting, Temporal…

Audio Jailbreak: An Open Comprehensive Benchmark for Jailbreaking Large Audio-Language Models
Evaluated models: SpeechGPT, SALMONN, DiVA+4 more
  • SpeechGPT
  • SALMONN
  • DiVA
  • Qwen 2 Audio
  • Llama Omni
  • Gemini 2.0 Flash
  • GPT-4o Audio

Code-Mixed Phonetic Attack

Published Analyzed Source: arXiv

A vulnerability exists in multiple large language and multimodal models that allows for the bypass of safety filters through the use of code-mixed prompts with phonetic perturbations. An attacker can craft a prompt in a code-mixed language (e.g., Hinglish) and apply phonetic misspellings to sensitive keywords (e.g., spelling "hate" as "haet"). This technique causes the model's tokenizer to parse the sensitive word into benign sub-tokens, preventing safety mechanisms from flagging the harmful…

" Haet Bhasha aur Diskrimineshun": Phonetic Perturbations in Code-Mixed Hinglish to Red-Team LLMs
Evaluated models: Gemma 1.1 7B IT, GPT-4o, GPT-4o Mini+2 more
  • Gemma 1.1 7B IT
  • GPT-4o
  • GPT-4o Mini
  • Llama 3 8B Instruct
  • Mistral 7B Instruct v0.3

Fine-Grained Speech LLM Control

Published Analyzed Source: arXiv

Speech-LLMs Qwen2-Audio (7B-Instruct) and Granite-Speech (3.2-8b) are vulnerable to universal acoustic adversarial attacks. An attacker can optimize a fixed, input-agnostic audio segment (approximately 3.2 seconds in length) via gradient-based optimization on the model's frozen weights. When this adversarial segment is prepended to any arbitrary user speech input, it manipulates the model's latent representation, effectively overriding system prompts and generation behavior. This vulnerability…

Universal Acoustic Adversarial Attacks for Flexible Control of Speech-LLMs
Evaluated models: Qwen 2 7B

On-Device LLM Hijacking

Published Analyzed Source: arXiv

Mobile LLM agents utilizing vision-based screen perception (OCR or Multimodal Large Language Models) are vulnerable to Visual Prompt Injection via malicious GUI overlays. An attacker holding the SYSTEM_ALERT_WINDOW permission can deploy non-focusable floating windows (using FLAG_NOT_FOCUSABLE) containing adversarial text or fabricated UI elements over legitimate applications. Because the agent captures the entire screen buffer to interpret the device state, it ingests the adversarial overlay…

From Assistants to Adversaries: Exploring the Security Risks of Mobile LLM Agents
Evaluated models: GPT-4o

SLM Imperceptible Noise Jailbreak

Published Analyzed Source: arXiv

An adversarial audio perturbation vulnerability exists in open-source Speech Language Models (SLMs), specifically Qwen2-Audio-7B-Instruct and LLaMa-Omni. The vulnerability allows remote attackers to bypass safety alignment mechanisms and jailbreak the model by injecting imperceptible adversarial noise into audio prompts. By utilizing white-box access and Projected Gradient Descent (PGD) optimization, an attacker can manipulate the continuous speech signal to trigger harmful responses (e.g…

SPIRIT: Patching Speech Language Models against Jailbreak Attacks
Evaluated models: Qwen 2 7B

LLM Self-Introspection Jailbreak

Published Analyzed Source: arXiv

A vulnerability exists in Large Language Models (LLMs) that allows attackers to manipulate the model's output by modifying token log probabilities. Attackers can use a lightweight plug-in model (BiasNet) to subtly alter the probabilities, steering the LLM toward generating harmful content even when safety mechanisms are in place. This attack requires only access to the top-k token log probabilities returned by the LLM's API, without needing model weights or internal access.

JULI: Jailbreak Large Language Models by Self-Introspection
Evaluated models: Llama 2 7B Chat, Llama 3 8B, Llama 3 8B Instruct+3 more
  • Llama 2 7B Chat
  • Llama 3 8B
  • Llama 3 8B Instruct
  • Mistral 7B
  • Qwen 2 1.5B Instruct
  • Qwen 2.5 1.5B Instruct

Latent-Space Jailbreak Optimization

Published Analyzed Source: arXiv

The LARGO attack exploits a vulnerability in Large Language Models (LLMs) allowing attackers to bypass safety mechanisms through the generation of "stealthy" adversarial prompts. The attack leverages gradient optimization in the LLM's continuous latent space to craft seemingly innocuous natural language suffixes which, when appended to harmful prompts, elicit unsafe responses. The vulnerability stems from the LLM's inability to reliably distinguish between benign and maliciously crafted latent…

LARGO: Latent Adversarial Reflection through Gradient Optimization for Jailbreaking LLMs
Evaluated models: Llama 2 13B Chat, Llama 2 7B Chat, Phi 3 Mini+1 more
  • Llama 2 13B Chat
  • Llama 2 7B Chat
  • Phi 3 Mini
  • Qwen 2.5 14B

Gradient-Based Privacy Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs) are vulnerable to a novel privacy jailbreak attack, dubbed PIG (Privacy Jailbreak Attack on LLMs via Gradient-based Iterative In-Context Optimization). PIG leverages in-context learning and gradient-based iterative optimization to extract Personally Identifiable Information (PII) from LLMs, bypassing built-in safety mechanisms. The attack iteratively refines a crafted prompt based on gradient information, focusing on tokens related to PII entities, thereby…

PIG: Privacy Jailbreak Attack on LLMs via Gradient-based Iterative In-Context Optimization
Evaluated models: Claude 3.5 Sonnet, GPT-4o, Llama 2 7B Chat+3 more
  • Claude 3.5 Sonnet
  • GPT-4o
  • Llama 2 7B Chat
  • Llama 3 8B Instruct
  • Mistral 7B Instruct v0.3
  • Vicuna 7B v1.5

Adversarial Suffix Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs), specifically instruction-tuned variants, are vulnerable to safety guardrail bypass via adversarial suffix injection. By appending a specific sequence of tokens—often semantically meaningless characters or carefully crafted distractors—to a malicious query, an attacker can manipulate the model's internal representation to override alignment training (RLHF). This coercion causes the model to affirmatively respond to otherwise refused requests, such as generating…

Adversarial Suffix Filtering: a Defense Pipeline for LLMs
Evaluated models: GPT-3.5, GPT-4o, Llama 2 7B+2 more
  • GPT-3.5
  • GPT-4o
  • Llama 2 7B
  • Llama 3.1 8B
  • Mistral 7B

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.