paper-with-me

Papers

Obfuscated Activations Bypass LLM Latent-Space Defenses

2024-12-12 · Luke Bailey, Alex Serrano, Abhay Sheshadri, Mikhail Seleznyov, Jordan Taylor, Erik Jenner, Jacob Hilton, Stephen Casper, Carlos Guestrin, Scott Emmons

Recent latent-space monitoring techniques have shown promise as defenses against LLM attacks. These defenses act as scanners that seek to detect harmful activations before they lead to undesirable actions. This prompts the question: Can models execute harmful behavior via inconspicuous latent states? Here, we study such obfuscated activations. We show that state-of-the-art latent-space defenses -- including sparse autoencoders, representation probing, and latent OOD detection -- are all vulnerable to obfuscated activations. For example, against probes trained to classify harmfulness, our attacks can often reduce recall from 100% to 0% while retaining a 90% jailbreaking rate. However, obfuscation has limits: we find that on a complex task (writing SQL code), obfuscation reduces model performance. Together, our results demonstrate that neural activations are highly malleable: we can reshape activation patterns in a variety of ways, often while preserving a network's behavior. This poses a fundamental challenge to latent-space defenses.

📄 PDF Abstract BibTeX arXiv:2412.09565

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Probing Latent Subspaces in LLM for AI Security: Identifying and Manipulating Adversarial States

2025-03-12 · Xin Wei Chia, Jonathan Pan

Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks, yet they remain vulnerable to adversarial manipulations such as jailbreaking via prompt injection attacks. These attacks bypass…

Dimensionality Reduction

Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples

2018-02-01 · ICML 2018 7 · Anish Athalye, Nicholas Carlini, David Wagner

We identify obfuscated gradients, a kind of gradient masking, as a phenomenon that leads to a false sense of security in defenses against adversarial examples. While defenses that cause obfuscated gradients appear to def…

Adversarial AttackAdversarial Defense

Bypassing Copyright Protection in Diffusion-based Customization via Two-Stage Latent Feature Optimization

2026-06-06 · Ziang Xu, Wenbo Yu, Hongyao Yu, Hao Fang 외 arxiv

With the growing concerns over copyright infringement in diffusion-based customization, adversarial attacks have emerged as a prominent defense strategy to prevent malicious content forgery in personalized image generati…

Personalized Image Generation

Encryption Inspired Adversarial Defense for Visual Classification

2020-05-16 · MaungMaung AprilPyone, Hitoshi Kiya

Conventional adversarial defenses reduce classification accuracy whether or not a model is under attacks. Moreover, most of image processing based defenses are defeated due to the problem of obfuscated gradients. In this…

Adversarial DefenseClassificationGeneral Classification

DRL-CLBA: A Clean Label Backdoor Attack for Speech Classification via DDPG Reinforcement Learning

2026-07-02 · Yueming Huang, Wenhan Yao, Fen Xiao, Xiarun Chen 외 arxiv

Deep learning models for speech classification are vulnerable to backdoor attacks, where malicious triggers cause misclassification at inference time. While sample-specific attacks can bypass many defenses, they often re…

Reinforcement Learning