Language Model Security Database

CRI (Compliance Refusal Initialization) initializes jailbreak attacks by leveraging pre-trained jailbreak prompts, effectively guiding the optimization process towards the compliance subspace of harmful prompts. This significantly enhances the success rate and reduces the computational overhead of attacks, often requiring only a single optimization step to bypass safety mechanisms. Attacks utilizing CRI demonstrate significantly improved ASR (Adversarial Success Rate) and reduced median steps…

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.

A vulnerability exists in Large Language Model (LLM) routing systems (control planes) that allows for the manipulation of inference flow via adversarial input sequences. LLM routers, which dynamically direct user queries to either "weak" (cheaper) or "strong" (expensive) models based on predicted query complexity, can be bypassed by appending specific, pre-optimized token sequences known as "confounder gadgets." These gadgets artificially inflate the router's complexity score for an input…

This vulnerability allows an attacker to bypass the safety mechanisms of Large Language Models (LLMs) by using an evolutionary algorithm to generate effective jailbreak prompts. The algorithm leverages the LLM's capabilities to iteratively refine prompts, increasing the likelihood of eliciting harmful responses to otherwise disallowed queries.

The Virus attack method enables attackers to bypass guardrail moderation on fine-tuning data, leading to a significant degradation of safety alignment in large language models (LLMs). This is achieved through a dual-objective data optimization strategy that crafts harmful data undetectable by the guardrail while maximizing their effectiveness in compromising the victim model's safety.

Large Language Models (LLMs), specifically instruction-following models using standard refusal training and adversarial training (such as Llama-3.1-8B-Instruct and Mistral-7B-V0.2), contain a vulnerability related to safety alignment bypass. The vulnerability arises from the models' inability to generalize safety reasoning to Out-Of-Distribution (OOD) inputs and scenarios involving competing objectives. Attackers can exploit this by employing linguistic manipulation (slang, uncommon dialects…

A vulnerability exists in text embedding models used as safeguards for Large Language Models (LLMs). Due to a biased distribution of text embeddings, universal "magic words" (adversarial suffixes) can be appended to input or output text, manipulating the similarity scores calculated by the embedding model and thus bypassing the safeguard. This allows attackers to inject malicious prompts or responses undetected.

Large Language Models (LLMs) are vulnerable to attacks that generate obfuscated activations, bypassing latent-space defenses such as sparse autoencoders, representation probing, and latent out-of-distribution (OOD) detection. Attackers can manipulate model inputs or training data to produce outputs exhibiting malicious behavior while remaining undetected by these defenses. This occurs because the models can represent harmful behavior through diverse activation patterns, allowing attackers to…

Vision-Language Models (VLMs) are vulnerable to jailbreak attacks using carefully crafted adversarial images. Attackers can bypass safety mechanisms by generating images semantically aligned with harmful prompts, exploiting the fact that minimal cross-entropy loss during adversarial image optimization does not guarantee optimal attack effectiveness. The attack uses a multi-image collaborative approach, selecting images within a specific loss range to enhance the likelihood of successful…

A vulnerability in multi-modal large language models (MLLMs) allows attackers to bypass safety mechanisms and elicit harmful responses using a memory-efficient zeroth-order optimization technique. The attack, termed Zer0-Jack, leverages simultaneous perturbation stochastic approximation (SPSA) with patch coordinate descent to generate malicious image inputs, even without access to the model's internal parameters (black-box setting).