Related papers: TrapSuffix: Proactive Defense Against Adversarial …
Recent studies have shown that large language models (LLMs) can infer private user attributes (e.g., age, location, gender) from user-generated text shared online, enabling rapid and large-scale privacy breaches. Existing…
Transferable backdoors pose a severe threat to the Pre-trained Language Models (PLMs) supply chain, yet defensive research remains nascent, primarily relying on detecting anomalies in the output feature space. We identify a critical flaw…
Decompositional jailbreaks pose a critical threat to large language models (LLMs) by allowing adversaries to fragment a malicious objective into a sequence of individually benign queries that collectively reconstruct prohibited content. In…
Despite the intrinsic risk-awareness of Large Language Models (LLMs), current defenses often result in shallow safety alignment, rendering models vulnerable to disguised attacks (e.g., prefilling) while degrading utility. To bridge this…
Although Large Language Models (LLMs) have demonstrated significant capabilities in executing complex tasks in a zero-shot manner, they are susceptible to jailbreak attacks and can be manipulated to produce harmful outputs. Recently, a…
Deep convolutional neural networks are susceptible to adversarial attacks. They can be easily deceived to give an incorrect output by adding a tiny perturbation to the input. This presents a great challenge in making CNNs robust against…
Jailbreak attacks represent one of the most sophisticated threats to the security of large language models (LLMs). To deal with such risks, we introduce an innovative framework that can help evaluate the effectiveness of jailbreak attacks…
As the use of large language models (LLMs) continues to expand, ensuring their safety and robustness has become a critical challenge. In particular, jailbreak attacks that bypass built-in safety mechanisms are increasingly recognized as a…
Recent advances in Large Language Models (LLMs) have led to impressive alignment where models learn to distinguish harmful from harmless queries through supervised finetuning (SFT) and reinforcement learning from human feedback (RLHF). In…
Existing white-box jailbreak attacks against aligned LLMs typically append discrete adversarial suffixes to the user prompt, which visibly alters the prompt and operates in a combinatorial token space. Prior work has avoided directly…
Jailbreaking techniques pose a significant threat to the safety of Large Language Models (LLMs). Existing defenses typically focus on single-turn attacks, lack coverage across languages, and rely on limited taxonomies that either fail to…
Large Language Models (LLMs) are increasingly integrated into real-world applications, from virtual assistants to autonomous agents. However, their flexibility also introduces new attack vectors-particularly Prompt Injection (PI), where…
Prompt injection attacks pose significant risks to language model safety, yet existing defenses are typically evaluated using classification performance. We show that high detection performance does not imply representational robustness.…
Ensuring safety alignment is a critical requirement for large language models (LLMs), particularly given increasing deployment in real-world applications. Despite considerable advancements, LLMs remain susceptible to jailbreak attacks,…
LLMs have made impressive progress, but their growing capabilities also expose them to highly flexible jailbreaking attacks designed to bypass safety alignment. While many existing defenses focus on known types of attacks, it is more…
The suffix array is an efficient data structure for in-memory pattern search. Suffix arrays can also be used for external-memory pattern search, via two-level structures that use an internal index to identify the correct block of suffix…
It is well known that adversarial attacks can fool deep neural networks with imperceptible perturbations. Although adversarial training significantly improves model robustness, failure cases of defense still broadly exist. In this work, we…
The increasing sophistication of large vision-language models (LVLMs) has been accompanied by advances in safety alignment mechanisms designed to prevent harmful content generation. However, these defenses remain vulnerable to sophisticated…
Backdoor attacks pose a critical threat by embedding hidden triggers into inputs, causing models to misclassify them into target labels. While extensive research has focused on mitigating these attacks in object recognition models through…
Website Fingerprinting attacks enable a passive eavesdropper to recover the user's otherwise anonymized web browsing activity by matching the observed traffic with prerecorded web traffic templates. The defenses that have been proposed to…