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.

VLM Split-Image Blindspot

Published Analyzed Source: arXiv

Vision Language Models (VLMs) utilizing independent vision encoders (e.g., ViT) and Large Language Model (LLM) decoders are vulnerable to Split-Image Visual Jailbreak Attacks (SIVA). The vulnerability arises from an architectural and alignment discrepancy: while the vision encoder processes image fragments (splits) in isolation via constrained attention or block-diagonal masks, the LLM decoder aggregates these features via cross-attention to reconstruct the semantic content. Current safety…

Robustness of Vision Language Models Against Split-Image Harmful Input Attacks
Evaluated models: Llama 3.2 11B

LLM Safety Geometry Fragility

Published Analyzed Source: arXiv

Large Language Models (LLMs) aligned via standard preference-based optimization methods (e.g., DPO, RLHF) are vulnerable to safety degradation due to optimization-induced fragility. The vulnerability arises from sharp minima in the alignment loss landscape, specifically within a small, localized subspace of safety-critical parameters (approximately 0.5% of neurons account for >80% of worst-case alignment loss). Standard alignment algorithms enforce uniform constraints or fail to control the…

Revisiting Robustness for LLM Safety Alignment via Selective Geometry Control
Evaluated models: Llama 3 8B, Llama 3.2 3B, Qwen 2.5 7B

Voice Agent Behavioral Bypass

Published Analyzed Source: arXiv

Audio Large Language Models (ALLMs) integrated into voice agent systems for high-stakes domains (banking, IT support, logistics) are vulnerable to multimodal adversarial attacks via spoken interaction. Adversaries can exploit the model's inherent compliance and contextual awareness through multi-turn dialogue to bypass authentication safeguards, escalate privileges (e.g., unauthorized credit limit increases), exfiltrate sensitive Personally Identifiable Information (PII), and poison…

Aegis: Towards Governance, Integrity, and Security of AI Voice Agents
Evaluated models: GPT-4o, GPT-4o Mini, Gemini 1.5 Pro+4 more
  • GPT-4o
  • GPT-4o Mini
  • Gemini 1.5 Pro
  • Gemini 2.5 Flash
  • Gemini 2.5 Pro
  • Qwen2-Audio 7B
  • Qwen 2.5 Omni 7B

Clinical Prompt Injection Harm

Published Analyzed Source: arXiv

Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) systems deployed in clinical workflows are vulnerable to direct and indirect (RAG-mediated) medical prompt injection attacks. Attackers can embed malicious instructions within user queries or external retrieved documents (such as poisoned clinical guidelines or PDFs). By exploiting "authority framing" (e.g., formatting the payload as a clinical guideline update or an editor's note), the injections successfully bypass generic…

MPIB: A Benchmark for Medical Prompt Injection Attacks and Clinical Safety in LLMs
Evaluated models: Qwen 2.5 7B Instruct, Qwen 2.5 32B Instruct, Qwen 2.5 72B Instruct+10 more
  • Qwen 2.5 7B Instruct
  • Qwen 2.5 32B Instruct
  • Qwen 2.5 72B Instruct
  • Llama 3.1 8B Instruct
  • Llama 3.1 70B Instruct
  • Mixtral 8x7B
  • Mixtral 8x22B v0.1
  • MedGemma 4B
  • MedGemma 27B
  • Meditron 7B
  • Meditron 70B
  • BioMistral 7B
  • MMed-Llama 3 8B

History Reweighting Jailbreak

Published Analyzed Source: arXiv

Large Language Models (LLMs), specifically LLaMA-3.2-11B, Qwen3-14B, Gemma 3-12B, and GPT-oss-20B, are vulnerable to black-box jailbreaking attacks via history-guided reinforcement learning (RL). The vulnerability arises from the models' inability to detect adversarial intent when prompts are iteratively refined based on historical interaction signals. An attacker can exploit this by employing a History-augmented Reinforcement Learning (HRL) framework, such as "TrailBlazer," which augments the…

TrailBlazer: History-Guided Reinforcement Learning for Black-Box LLM Jailbreaking
Evaluated models: GPT-oss 20B, Llama 3.2 11B, Qwen 3 14B+1 more
  • GPT-oss 20B
  • Llama 3.2 11B
  • Qwen 3 14B
  • Gemma 3 12B

Shallow Alignment Jailbreak

Published Analyzed Source: arXiv

A vulnerability exists in aligned Large Language Models (LLMs) related to "shallow safety alignment," where safety mechanisms disproportionately rely on the initial tokens generated by the model. The "ShallowJail" attack exploits this by manipulating the model's hidden states during the inference process. Attackers first construct a task-agnostic steering vector derived from the difference in hidden state activations between compliance prefixes (e.g., "Sure, here are the details") and refusal…

ShallowJail: Steering Jailbreaks against Large Language Models
Evaluated models: Llama 3.1 8B, Qwen 2.5 7B

Causal Front-Door Jailbreak

Published Analyzed Source: arXiv

A vulnerability exists in safety-aligned Large Language Models (LLMs) wherein internal safety alignment mechanisms (such as RLHF) function as unobserved causal confounders rather than erasing prohibited knowledge. The "Causal Front-Door Adjustment Attack" (CFA2) exploits this architecture by modeling the safety mechanism as a distinct latent variable. Attackers with white-box access can employ Sparse Autoencoders (SAEs) to disentangle dense internal representations into sparse features…

Causal Front-Door Adjustment for Robust Jailbreak Attacks on LLMs
Evaluated models: Llama 2 7B, Llama 3.1 8B, Mistral 7B+1 more
  • Llama 2 7B
  • Llama 3.1 8B
  • Mistral 7B
  • Gemma 2 9B

Chunky Post-Training Miscalibration

Published Analyzed Source: arXiv

Large Language Models (LLMs) exhibit a vulnerability termed "chunky post-training," where the model learns spurious correlations between incidental prompt features (e.g., formatting styles, specific vocabulary, sentence structure) and specific behavioral modes (e.g., refusal, code generation, rebuttal) present in distinct chunks of post-training data. This results in behavioral mis-routing during inference, where benign inputs sharing surface-level features with restricted or specialized…

Chunky Post-Training: Data Driven Failures of Generalization
Evaluated models: Claude Haiku 4.5, Claude Sonnet 4.5, Claude Opus 4.5+5 more
  • Claude Haiku 4.5
  • Claude Sonnet 4.5
  • Claude Opus 4.5
  • GPT-5.1
  • Grok 4.1 Mini
  • Tülu 3 8B
  • Tülu 3 70B
  • Tülu 3 405B

Covert Template System Backdoor

Published Analyzed Source: arXiv

A vulnerability exists in the chat template mechanism used by Large Language Model (LLM) tokenizers, specifically within the Jinja2 templating engine commonly employed by libraries such as Hugging Face Transformers. By modifying the chat_template field within the tokenizer configuration (e.g., tokenizer_config.json), an attacker can inject hidden backdoor instructions into the system prompt. These instructions are concatenated with legitimate user inputs during the tokenization process…

BadTemplate: A Training-Free Backdoor Attack via Chat Template Against Large Language Models
Evaluated models: Llama 3.1 8B Instruct, DeepSeek LLM 7B Chat, Llama 3.1 70B Instruct+6 more
  • Llama 3.1 8B Instruct
  • DeepSeek LLM 7B Chat
  • Llama 3.1 70B Instruct
  • Yi 1.5 Chat 34B
  • Mistral Small Instruct 2409
  • Phi-3.5 Mini Instruct
  • DeepSeek V3.2 Exp
  • Gemini 2.5 Flash
  • GPT-4o Mini

Evolutionary Hidden Knowledge Recovery

Published Analyzed Source: arXiv

Large Language Models (LLMs) subjected to machine unlearning techniques (specifically AltPO, GradDiff, IDKDPO, IDKNLL, UNDIAL, NPO, and SimNPO) contain a vulnerability regarding the persistence of latent knowledge. Despite achieving high "forgetting" scores on standard, benign benchmarks, these models remain susceptible to black-box evolutionary adversarial attacks. An attacker can utilize an automated framework (REBEL) comprising a "Hacker" model and a "Judge" model to iteratively mutate…

REBEL: Hidden Knowledge Recovery via Evolutionary-Based Evaluation Loop
Evaluated models: Not reported

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.