Towards LLM Guardrails via Sparse Representation Steering

Large Language Models (LLMs) have demonstrated remarkable performance in natural language generation tasks, yet their uncontrolled outputs pose significant ethical and safety risks. Recently, representation engineering methods have shown promising results in steering model behavior by modifying the rich semantic information encoded in activation vectors. However, due to the difficulty of precisely disentangling semantic directions within high-dimensional representation space, existing approaches suffer from three major limitations: lack of fine-grained control, quality degradation of generated content, and poor interpretability. To address these challenges, we propose a sparse encoding-based representation engineering method, named SRE, which decomposes polysemantic activations into a structured, monosemantic feature space. By leveraging sparse autoencoding, our approach isolates and adjusts only task-specific sparse feature dimensions, enabling precise and interpretable steering of model behavior while preserving content quality. We validate our method on three critical domains, i.e., safety, fairness, and truthfulness using the open-source LLM Gemma-2-2B-it. Experimental results show that SRE achieves superior controllability while maintaining the overall quality of generated content (i.e., controllability and quality), demonstrating its effectiveness as a fine-grained and interpretable activation steering framework.

Harnessing Frequency Spectrum Insights for Image Copyright Protection Against Diffusion Models

Diffusion models have achieved remarkable success in novel view synthesis, but their reliance on large, diverse, and often untraceable Web datasets has raised pressing concerns about image copyright protection. Current methods fall short in reliably identifying unauthorized image use, as they struggle to generalize across varied generation tasks and fail when the training dataset includes images from multiple sources with few identifiable (watermarked or poisoned) samples. In this paper, we present novel evidence that diffusion-generated images faithfully preserve the statistical properties of their training data, particularly reflected in their spectral features. Leveraging this insight, we introduce \emph{CoprGuard}, a robust frequency domain watermarking framework to safeguard against unauthorized image usage in diffusion model training and fine-tuning. CoprGuard demonstrates remarkable effectiveness against a wide range of models, from naive diffusion models to sophisticated text-to-image models, and is robust even when watermarked images comprise a mere 1\% of the training dataset. This robust and versatile approach empowers content owners to protect their intellectual property in the era of AI-driven image generation.

Sparse Autoencoder as a Zero-Shot Classifier for Concept Erasing in Text-to-Image Diffusion Models

Text-to-image (T2I) diffusion models have achieved remarkable progress in generating high-quality images but also raise people's concerns about generating harmful or misleading content. While extensive approaches have been proposed to erase unwanted concepts without requiring retraining from scratch, they inadvertently degrade performance on normal generation tasks. In this work, we propose Interpret then Deactivate (ItD), a novel framework to enable precise concept removal in T2I diffusion models while preserving overall performance. ItD first employs a sparse autoencoder (SAE) to interpret each concept as a combination of multiple features. By permanently deactivating the specific features associated with target concepts, we repurpose SAE as a zero-shot classifier that identifies whether the input prompt includes target concepts, allowing selective concept erasure in diffusion models. Moreover, we demonstrate that ItD can be easily extended to erase multiple concepts without requiring further training. Comprehensive experiments across celebrity identities, artistic styles, and explicit content demonstrate ItD's effectiveness in eliminating targeted concepts without interfering with normal concept generation. Additionally, ItD is also robust against adversarial prompts designed to circumvent content filters. Code is available at: https://github.com/NANSirun/Interpret-then-deactivate.

Can Small Language Models Reliably Resist Jailbreak Attacks? A Comprehensive Evaluation

Small language models (SLMs) have emerged as promising alternatives to large language models (LLMs) due to their low computational demands, enhanced privacy guarantees and comparable performance in specific domains through light-weight fine-tuning. Deploying SLMs on edge devices, such as smartphones and smart vehicles, has become a growing trend. However, the security implications of SLMs have received less attention than LLMs, particularly regarding jailbreak attacks, which is recognized as one of the top threats of LLMs by the OWASP. In this paper, we conduct the first large-scale empirical study of SLMs' vulnerabilities to jailbreak attacks. Through systematically evaluation on 63 SLMs from 15 mainstream SLM families against 8 state-of-the-art jailbreak methods, we demonstrate that 47.6% of evaluated SLMs show high susceptibility to jailbreak attacks (ASR > 40%) and 38.1% of them can not even resist direct harmful query (ASR > 50%). We further analyze the reasons behind the vulnerabilities and identify four key factors: model size, model architecture, training datasets and training techniques. Moreover, we assess the effectiveness of three prompt-level defense methods and find that none of them achieve perfect performance, with detection accuracy varying across different SLMs and attack methods. Notably, we point out that the inherent security awareness play a critical role in SLM security, and models with strong security awareness could timely terminate unsafe response with little reminder. Building upon the findings, we highlight the urgent need for security-by-design approaches in SLM development and provide valuable insights for building more trustworthy SLM ecosystem.

Towards Collaborative Anti-Money Laundering Among Financial Institutions

Money laundering is the process that intends to legalize the income derived from illicit activities, thus facilitating their entry into the monetary flow of the economy without jeopardizing their source. It is crucial to identify such activities accurately and reliably in order to enforce anti-money laundering (AML). Despite considerable efforts to AML, a large number of such activities still go undetected. Rule-based methods were first introduced and are still widely used in current detection systems. With the rise of machine learning, graph-based learning methods have gained prominence in detecting illicit accounts through the analysis of money transfer graphs. Nevertheless, these methods generally assume that the transaction graph is centralized, whereas in practice, money laundering activities usually span multiple financial institutions. Due to regulatory, legal, commercial, and customer privacy concerns, institutions tend not to share data, restricting their utility in practical usage. In this paper, we propose the first algorithm that supports performing AML over multiple institutions while protecting the security and privacy of local data. To evaluate, we construct Alipay-ECB, a real-world dataset comprising digital transactions from Alipay, the world's largest mobile payment platform, alongside transactions from E-Commerce Bank (ECB). The dataset includes over 200 million accounts and 300 million transactions, covering both intra-institution transactions and those between Alipay and ECB. This makes it the largest real-world transaction graph available for analysis. The experimental results demonstrate that our methods can effectively identify cross-institution money laundering subgroups. Additionally, experiments on synthetic datasets also demonstrate that our method is efficient, requiring only a few minutes on datasets with millions of transactions.

Towards Label-Only Membership Inference Attack against Pre-trained Large Language Models

Membership Inference Attacks (MIAs) aim to predict whether a data sample belongs to the model's training set or not. Although prior research has extensively explored MIAs in Large Language Models (LLMs), they typically require accessing to complete output logits (\ie, \textit{logits-based attacks}), which are usually not available in practice. In this paper, we study the vulnerability of pre-trained LLMs to MIAs in the \textit{label-only setting}, where the adversary can only access generated tokens (text). We first reveal that existing label-only MIAs have minor effects in attacking pre-trained LLMs, although they are highly effective in inferring fine-tuning datasets used for personalized LLMs. We find that their failure stems from two main reasons, including better generalization and overly coarse perturbation. Specifically, due to the extensive pre-training corpora and exposing each sample only a few times, LLMs exhibit minimal robustness differences between members and non-members. This makes token-level perturbations too coarse to capture such differences. To alleviate these problems, we propose \textbf{PETAL}: a label-only membership inference attack based on \textbf{PE}r-\textbf{T}oken sem\textbf{A}ntic simi\textbf{L}arity. Specifically, PETAL leverages token-level semantic similarity to approximate output probabilities and subsequently calculate the perplexity. It finally exposes membership based on the common assumption that members are `better' memorized and have smaller perplexity. We conduct extensive experiments on the WikiMIA benchmark and the more challenging MIMIR benchmark. Empirically, our PETAL performs better than the extensions of existing label-only attacks against personalized LLMs and even on par with other advanced logit-based attacks across all metrics on five prevalent open-source LLMs.

REFINE: Inversion-Free Backdoor Defense via Model Reprogramming

Backdoor attacks on deep neural networks (DNNs) have emerged as a significant security threat, allowing adversaries to implant hidden malicious behaviors during the model training phase. Pre-processing-based defense, which is one of the most important defense paradigms, typically focuses on input transformations or backdoor trigger inversion (BTI) to deactivate or eliminate embedded backdoor triggers during the inference process. However, these methods suffer from inherent limitations: transformation-based defenses often fail to balance model utility and defense performance, while BTI-based defenses struggle to accurately reconstruct trigger patterns without prior knowledge. In this paper, we propose REFINE, an inversion-free backdoor defense method based on model reprogramming. REFINE consists of two key components: \textbf{(1)} an input transformation module that disrupts both benign and backdoor patterns, generating new benign features; and \textbf{(2)} an output remapping module that redefines the model's output domain to guide the input transformations effectively. By further integrating supervised contrastive loss, REFINE enhances the defense capabilities while maintaining model utility. Extensive experiments on various benchmark datasets demonstrate the effectiveness of our REFINE and its resistance to potential adaptive attacks.

CoKV: Optimizing KV Cache Allocation via Cooperative Game

Large language models (LLMs) have achieved remarkable success on various aspects of human life. However, one of the major challenges in deploying these models is the substantial memory consumption required to store key-value pairs (KV), which imposes significant resource demands. Recent research has focused on KV cache budget allocation, with several approaches proposing head-level budget distribution by evaluating the importance of individual attention heads. These methods, however, assess the importance of heads independently, overlooking their cooperative contributions within the model, which may result in a deviation from their true impact on model performance. In light of this limitation, we propose CoKV, a novel method that models the cooperation between heads in model inference as a cooperative game. By evaluating the contribution of each head within the cooperative game, CoKV can allocate the cache budget more effectively. Extensive experiments show that CoKV achieves state-of-the-art performance on the LongBench benchmark using LLama-3-8B-Instruct and Mistral-7B models.

Robust Watermarks Leak: Channel-Aware Feature Extraction Enables Adversarial Watermark Manipulation

Watermarking plays a key role in the provenance and detection of AI-generated content. While existing methods prioritize robustness against real-world distortions (e.g., JPEG compression and noise addition), we reveal a fundamental tradeoff: such robust watermarks inherently improve the redundancy of detectable patterns encoded into images, creating exploitable information leakage. To leverage this, we propose an attack framework that extracts leakage of watermark patterns through multi-channel feature learning using a pre-trained vision model. Unlike prior works requiring massive data or detector access, our method achieves both forgery and detection evasion with a single watermarked image. Extensive experiments demonstrate that our method achieves a 60\% success rate gain in detection evasion and 51\% improvement in forgery accuracy compared to state-of-the-art methods while maintaining visual fidelity. Our work exposes the robustness-stealthiness paradox: current "robust" watermarks sacrifice security for distortion resistance, providing insights for future watermark design.

Activation Approximations Can Incur Safety Vulnerabilities Even in Aligned LLMs: Comprehensive Analysis and Defense

Large Language Models (LLMs) have showcased remarkable capabilities across various domains. Accompanying the evolving capabilities and expanding deployment scenarios of LLMs, their deployment challenges escalate due to their sheer scale and the advanced yet complex activation designs prevalent in notable model series, such as Llama, Gemma, and Mistral. These challenges have become particularly pronounced in resource-constrained deployment scenarios, where mitigating inference efficiency bottlenecks is imperative. Among various recent efforts, activation approximation has emerged as a promising avenue for pursuing inference efficiency, sometimes considered indispensable in applications such as private inference. Despite achieving substantial speedups with minimal impact on utility, even appearing sound and practical for real-world deployment, the safety implications of activation approximations remain unclear. In this work, we fill this critical gap in LLM safety by conducting the first systematic safety evaluation of activation approximations. Our safety vetting spans seven sota techniques across three popular categories, revealing consistent safety degradation across ten safety-aligned LLMs.