Related papers: A Single Neuron Is Sufficient to Bypass Safety Ali…

Prior work argues that refusal in large language models is mediated by a single activation-space direction, enabling effective steering and ablation. We show that this account is incomplete. Across eleven categories of refusal and…

Safety alignment in large language models (LLMs), particularly for cybersecurity tasks, primarily focuses on preventing misuse. While this approach reduces direct harm, it obscures a complementary failure mode: denial of assistance to…

Perturbation probing generates task-specific causal hypotheses for FFN neurons in large language models using two forward passes per prompt and no backpropagation, followed by a one-time intervention sweep of about 150 passes amortized…

Large Language Models (LLMs) are increasingly attracting attention in various applications. Nonetheless, there is a growing concern as some users attempt to exploit these models for malicious purposes, including the synthesis of controlled…

The safety alignment of large language models (LLMs) can be circumvented through adversarially crafted inputs, yet the mechanisms by which these attacks bypass safety barriers remain poorly understood. Prior work suggests that a single…

Multi-modal large language models (MLLMs) have made significant progress, yet their safety alignment remains limited. Typically, current open-source MLLMs rely on the alignment inherited from their language module to avoid harmful…

The emergence of finetuning-as-a-service has revealed a new vulnerability in large language models (LLMs). A mere handful of malicious data uploaded by users can subtly manipulate the finetuning process, resulting in an alignment-broken…

With the growing accessibility and wide adoption of large language models, concerns about their safety and alignment with human values have become paramount. In this paper, we identify a concerning phenomenon: Reasoning-Induced Misalignment…

Large Language Models (LLMs) are vulnerable to jailbreak attacks that exploit weaknesses in traditional safety alignment, which often relies on rigid refusal heuristics or representation engineering to block harmful outputs. While they are…

Despite investments in improving model safety, studies show that misaligned capabilities remain latent in safety-tuned models. In this work, we shed light on the mechanics of this phenomenon. First, we show that even when model generations…

Despite vast research in adversarial examples, the root causes of model susceptibility are not well understood. Instead of looking at attack-specific robustness, we propose a notion that evaluates the sensitivity of individual neurons in…

Despite significant progress in safety alignment, large language models (LLMs) remain susceptible to jailbreak attacks. Existing defense mechanisms have not fully deleted harmful knowledge in LLMs, which allows such attacks to bypass…

Safety-aligned language models often refuse cybersecurity requests whose wording resembles misuse, even when the task is authorized and defensive. This makes security evaluation ambiguous: a failed answer may reflect missing capability or…

Hallucination in large language models (LLMs) has been widely studied in recent years, with progress in both detection and mitigation aimed at improving truthfulness. Yet, a critical side effect remains largely overlooked: enhancing…

Safety alignment is essential for the responsible deployment of large language models (LLMs). Yet, existing approaches often rely on heavyweight fine-tuning that is costly to update, audit, and maintain across model families. Full…

Instruction-following language models often show undesirable biases. These undesirable biases may be accelerated in the real-world usage of language models, where a wide range of instructions is used through zero-shot example prompting. To…

Safety alignment in large language models (LLMs) is primarily evaluated under open-ended generation, where models can mitigate risk by refusing to respond. In contrast, many real-world applications place LLMs in structured decision-making…

Pretrained language models have been shown to significantly predict brain recordings of people comprehending language. Recent work suggests that the prediction of the next word is a key mechanism that contributes to this alignment. What is…

In real-world deployments, Vision-Language Large Models (VLLMs) face critical challenges from multilingual and multimodal composite attacks: harmful images paired with low-resource language texts can easily bypass defenses designed for…

Large language models (LLMs) are typically aligned to refuse harmful instructions through safety fine-tuning. A recent attack, termed abliteration, identifies and suppresses the single latent direction most responsible for refusal behavior,…