arXiv subjects

Xingjun Ma

Publications and source records attributed to Xingjun Ma.

At least 109 records · Page 6Linked to original sources

arXiv2025

GenBreak: Red Teaming Text-to-Image Generators Using Large Language Models

Text-to-image (T2I) models such as Stable Diffusion have advanced rapidly and are now widely used in content creation. However, these models can be misused to generate harmful content, including nudity or violence, posing significant safety risks. While most platforms employ content moderation systems, underlying vulnerabilities can still be exploited by determined adversaries. Recent research on red-teaming and adversarial attacks against T2I models has notable limitations: some studies successfully generate highly toxic images but use adversarial prompts that are easily detected and blocked by safety filters, while others focus on bypassing safety mechanisms but fail to produce genuinely harmful outputs, neglecting the discovery of truly high-risk prompts. Consequently, there remains a lack of reliable tools for evaluating the safety of defended T2I models. To address this gap, we propose GenBreak, a framework that fine-tunes a red-team large language model (LLM) to systematically explore underlying vulnerabilities in T2I generators. Our approach combines supervised fine-tuning on curated datasets with reinforcement learning via interaction with a surrogate T2I model. By integrating multiple reward signals, we guide the LLM to craft adversarial prompts that enhance both evasion capability and image toxicity, while maintaining semantic coherence and diversity. These prompts demonstrate strong effectiveness in black-box attacks against commercial T2I generators, revealing practical and concerning safety weaknesses.

cs.CR↗

arXiv2025

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment

The remarkable success of Large Language Models (LLMs) has illuminated a promising pathway toward achieving Artificial General Intelligence for both academic and industrial communities, owing to their unprecedented performance across various applications. As LLMs continue to gain prominence in both research and commercial domains, their security and safety implications have become a growing concern, not only for researchers and corporations but also for every nation. Currently, existing surveys on LLM safety primarily focus on specific stages of the LLM lifecycle, e.g., deployment phase or fine-tuning phase, lacking a comprehensive understanding of the entire "lifechain" of LLMs. To address this gap, this paper introduces, for the first time, the concept of "full-stack" safety to systematically consider safety issues throughout the entire process of LLM training, deployment, and eventual commercialization. Compared to the off-the-shelf LLM safety surveys, our work demonstrates several distinctive advantages: (I) Comprehensive Perspective. We define the complete LLM lifecycle as encompassing data preparation, pre-training, post-training, deployment and final commercialization. To our knowledge, this represents the first safety survey to encompass the entire lifecycle of LLMs. (II) Extensive Literature Support. Our research is grounded in an exhaustive review of over 800+ papers, ensuring comprehensive coverage and systematic organization of security issues within a more holistic understanding. (III) Unique Insights. Through systematic literature analysis, we have developed reliable roadmaps and perspectives for each chapter. Our work identifies promising research directions, including safety in data generation, alignment techniques, model editing, and LLM-based agent systems. These insights provide valuable guidance for researchers pursuing future work in this field.

cs.CR↗

arXiv2025

RedRFT: A Light-Weight Benchmark for Reinforcement Fine-Tuning-Based Red Teaming

Red teaming has proven to be an effective method for identifying and mitigating vulnerabilities in Large Language Models (LLMs). Reinforcement Fine-Tuning (RFT) has emerged as a promising strategy among existing red teaming techniques. However, a lack of a unified benchmark hinders current RFT-based red teaming methods. Implementation details, especially in Proximal Policy Optimization (PPO)-based RFT, significantly affect outcome stability and reproducibility. To address this issue, we introduce RedRFT, a lightweight benchmark designed to simplify and standardize the implementation and evaluation of RFT-based red teaming. RedRFT combines the design strengths of both single-file CleanRL and highly modularized Tianshou, offering high-quality single-file red teaming implementations and modular PPO core components, such as the General Advantage Estimator. It supports a variety of token and sentence diversity metrics, featuring modularized intrinsic reward computation that facilitates plug-and-play experimentation. To clarify their influence on RFT performance, we conducted an extensive ablation study on key components, including Low-Rank Adaptation (LoRA), Kullback-Leibler (KL) divergence, and Lagrange Multiplier. We hope this work contributes to 1) gaining a comprehensive understanding of the implementation nuances of RFT-based red teaming algorithms, and 2) enabling rapid prototyping of innovative features for RFT-based red teaming. Code for the benchmark can be accessed at https://github.com/x-zheng16/RedRFT.git.

cs.LG↗

arXiv2025

X-Transfer Attacks: Towards Super Transferable Adversarial Attacks on CLIP

As Contrastive Language-Image Pre-training (CLIP) models are increasingly adopted for diverse downstream tasks and integrated into large vision-language models (VLMs), their susceptibility to adversarial perturbations has emerged as a critical concern. In this work, we introduce \textbf{X-Transfer}, a novel attack method that exposes a universal adversarial vulnerability in CLIP. X-Transfer generates a Universal Adversarial Perturbation (UAP) capable of deceiving various CLIP encoders and downstream VLMs across different samples, tasks, and domains. We refer to this property as \textbf{super transferability}--a single perturbation achieving cross-data, cross-domain, cross-model, and cross-task adversarial transferability simultaneously. This is achieved through \textbf{surrogate scaling}, a key innovation of our approach. Unlike existing methods that rely on fixed surrogate models, which are computationally intensive to scale, X-Transfer employs an efficient surrogate scaling strategy that dynamically selects a small subset of suitable surrogates from a large search space. Extensive evaluations demonstrate that X-Transfer significantly outperforms previous state-of-the-art UAP methods, establishing a new benchmark for adversarial transferability across CLIP models. The code is publicly available in our \href{https://github.com/HanxunH/XTransferBench}{GitHub repository}.

cs.CV↗

arXiv2025

From Failures to Fixes: LLM-Driven Scenario Repair for Self-Evolving Autonomous Driving

Ensuring robust and generalizable autonomous driving requires not only broad scenario coverage but also efficient repair of failure cases, particularly those related to challenging and safety-critical scenarios. However, existing scenario generation and selection methods often lack adaptivity and semantic relevance, limiting their impact on performance improvement. In this paper, we propose \textbf{SERA}, an LLM-powered framework that enables autonomous driving systems to self-evolve by repairing failure cases through targeted scenario recommendation. By analyzing performance logs, SERA identifies failure patterns and dynamically retrieves semantically aligned scenarios from a structured bank. An LLM-based reflection mechanism further refines these recommendations to maximize relevance and diversity. The selected scenarios are used for few-shot fine-tuning, enabling targeted adaptation with minimal data. Experiments on the benchmark show that SERA consistently improves key metrics across multiple autonomous driving baselines, demonstrating its effectiveness and generalizability under safety-critical conditions.

cs.CV↗

arXiv2025

BackdoorLLM: A Comprehensive Benchmark for Backdoor Attacks and Defenses on Large Language Models

Generative large language models (LLMs) have achieved state-of-the-art results on a wide range of tasks, yet they remain susceptible to backdoor attacks: carefully crafted triggers in the input can manipulate the model to produce adversary-specified outputs. While prior research has predominantly focused on backdoor risks in vision and classification settings, the vulnerability of LLMs in open-ended text generation remains underexplored. To fill this gap, we introduce BackdoorLLM (Our BackdoorLLM benchmark was awarded First Prize in the SafetyBench competition, https://www.mlsafety.org/safebench/winners, organized by the Center for AI Safety, https://safe.ai/.), the first comprehensive benchmark for systematically evaluating backdoor threats in text-generation LLMs. BackdoorLLM provides: (i) a unified repository of benchmarks with a standardized training and evaluation pipeline; (ii) a diverse suite of attack modalities, including data poisoning, weight poisoning, hidden-state manipulation, and chain-of-thought hijacking; (iii) over 200 experiments spanning 8 distinct attack strategies, 7 real-world scenarios, and 6 model architectures; (iv) key insights into the factors that govern backdoor effectiveness and failure modes in LLMs; and (v) a defense toolkit encompassing 7 representative mitigation techniques. Our code and datasets are available at https://github.com/bboylyg/BackdoorLLM. We will continuously incorporate emerging attack and defense methodologies to support the research in advancing the safety and reliability of LLMs.

cs.AI↗

arXiv2025

SafeVid: Toward Safety Aligned Video Large Multimodal Models

As Video Large Multimodal Models (VLMMs) rapidly advance, their inherent complexity introduces significant safety challenges, particularly the issue of mismatched generalization where static safety alignments fail to transfer to dynamic video contexts. We introduce SafeVid, a framework designed to instill video-specific safety principles in VLMMs. SafeVid uniquely transfers robust textual safety alignment capabilities to the video domain by employing detailed textual video descriptions as an interpretive bridge, facilitating LLM-based rule-driven safety reasoning. This is achieved through a closed-loop system comprising: 1) generation of SafeVid-350K, a novel 350,000-pair video-specific safety preference dataset; 2) targeted alignment of VLMMs using Direct Preference Optimization (DPO); and 3) comprehensive evaluation via our new SafeVidBench benchmark. Alignment with SafeVid-350K significantly enhances VLMM safety, with models like LLaVA-NeXT-Video demonstrating substantial improvements (e.g., up to 42.39%) on SafeVidBench. SafeVid provides critical resources and a structured approach, demonstrating that leveraging textual descriptions as a conduit for safety reasoning markedly improves the safety alignment of VLMMs. We have made SafeVid-350K dataset (https://huggingface.co/datasets/yxwang/SafeVid-350K) publicly available.

cs.CV↗

arXiv2025

TokenMark: A Modality-Agnostic Watermark for Pre-trained Transformers

Watermarking is a critical tool for model ownership verification. However, existing watermarking techniques are often designed for specific data modalities and downstream tasks, without considering the inherent architectural properties of the model. This lack of generality and robustness underscores the need for a more versatile watermarking approach. In this work, we investigate the properties of Transformer models and propose TokenMark, a modality-agnostic, robust watermarking system for pre-trained models, leveraging the permutation equivariance property. TokenMark embeds the watermark by fine-tuning the pre-trained model on a set of specifically permuted data samples, resulting in a watermarked model that contains two distinct sets of weights -- one for normal functionality and the other for watermark extraction, the latter triggered only by permuted inputs. Extensive experiments on state-of-the-art pre-trained models demonstrate that TokenMark significantly improves the robustness, efficiency, and universality of model watermarking, highlighting its potential as a unified watermarking solution.

cs.CR↗

arXiv2025

Toward Generalizable Evaluation in the LLM Era: A Survey Beyond Benchmarks

Large Language Models (LLMs) are advancing at an amazing speed and have become indispensable across academia, industry, and daily applications. To keep pace with the status quo, this survey probes the core challenges that the rise of LLMs poses for evaluation. We identify and analyze two pivotal transitions: (i) from task-specific to capability-based evaluation, which reorganizes benchmarks around core competencies such as knowledge, reasoning, instruction following, multi-modal understanding, and safety; and (ii) from manual to automated evaluation, encompassing dynamic dataset curation and "LLM-as-a-judge" scoring. Yet, even with these transitions, a crucial obstacle persists: the evaluation generalization issue. Bounded test sets cannot scale alongside models whose abilities grow seemingly without limit. We will dissect this issue, along with the core challenges of the above two transitions, from the perspectives of methods, datasets, evaluators, and metrics. Due to the fast evolving of this field, we will maintain a living GitHub repository (links are in each section) to crowd-source updates and corrections, and warmly invite contributors and collaborators.

cs.CL↗

arXiv2025

CHASE: A Causal Hypergraph based Framework for Root Cause Analysis in Multimodal Microservice Systems

In recent years, the widespread adoption of distributed microservice architectures within the industry has significantly increased the demand for enhanced system availability and robustness. Due to the complex service invocation paths and dependencies in enterprise-level microservice systems, it is challenging to locate the anomalies promptly during service invocations, thus causing intractable issues for normal system operations and maintenance. In this paper, we propose a Causal Heterogeneous grAph baSed framEwork for root cause analysis, namely CHASE, for microservice systems with multimodal data, including traces, logs, and system monitoring metrics. Specifically, related information is encoded into representative embeddings and further modeled by a multimodal invocation graph. Following that, anomaly detection is performed on each instance node with attentive heterogeneous message passing from its adjacent metric and log nodes. Finally, CHASE learns from the constructed hypergraph with hyperedges representing the flow of causality and performs root cause localization. We evaluate the proposed framework on two public microservice datasets with distinct attributes and compare with the state-of-the-art methods. The results show that CHASE achieves the average performance gain up to 36.2%(A@1) and 29.4%(Percentage@1), respectively to its best counterpart.

cs.LG↗

arXiv2025

FedEGG: Federated Learning with Explicit Global Guidance

Federated Learning (FL) holds great potential for diverse applications owing to its privacy-preserving nature. However, its convergence is often challenged by non-IID data distributions, limiting its effectiveness in real-world deployments. Existing methods help address these challenges via optimization-based client constraints, adaptive client selection, or the use of pre-trained models or synthetic data. In this work, we reinterpret these approaches as all introducing an \emph{implicit guiding task} to regularize and steer client learning. Following this insight, we propose to introduce an \emph{explicit global guiding task} into the current FL framework to improve convergence and performance. To this end, we present \textbf{FedEGG}, a new FL algorithm that constructs a global guiding task using a well-defined, easy-to-converge learning task based on a public dataset and Large Language Models (LLMs). This approach effectively combines the strengths of federated (the original FL task) and centralized (the global guiding task) learning. We provide a theoretical analysis of FedEGG's convergence, examining the impact of data heterogeneity between the guiding and FL tasks and the guiding strength. Our analysis derives an upper bound for the optimal guiding strength, offering practical insights for implementation. Empirically, FedEGG demonstrates superior performance over state-of-the-art FL methods under both IID and non-IID settings, and further improves their performances when combined.

cs.LG↗

arXiv2025

AnyAttack: Towards Large-scale Self-supervised Adversarial Attacks on Vision-language Models

Due to their multimodal capabilities, Vision-Language Models (VLMs) have found numerous impactful applications in real-world scenarios. However, recent studies have revealed that VLMs are vulnerable to image-based adversarial attacks. Traditional targeted adversarial attacks require specific targets and labels, limiting their real-world impact.We present AnyAttack, a self-supervised framework that transcends the limitations of conventional attacks through a novel foundation model approach. By pre-training on the massive LAION-400M dataset without label supervision, AnyAttack achieves unprecedented flexibility - enabling any image to be transformed into an attack vector targeting any desired output across different VLMs.This approach fundamentally changes the threat landscape, making adversarial capabilities accessible at an unprecedented scale. Our extensive validation across five open-source VLMs (CLIP, BLIP, BLIP2, InstructBLIP, and MiniGPT-4) demonstrates AnyAttack's effectiveness across diverse multimodal tasks. Most concerning, AnyAttack seamlessly transfers to commercial systems including Google Gemini, Claude Sonnet, Microsoft Copilot and OpenAI GPT, revealing a systemic vulnerability requiring immediate attention.

cs.LG↗

arXiv2025

Towards Million-Scale Adversarial Robustness Evaluation With Stronger Individual Attacks

As deep learning models are increasingly deployed in safety-critical applications, evaluating their vulnerabilities to adversarial perturbations is essential for ensuring their reliability and trustworthiness. Over the past decade, a large number of white-box adversarial robustness evaluation methods (i.e., attacks) have been proposed, ranging from single-step to multi-step methods and from individual to ensemble methods. Despite these advances, challenges remain in conducting meaningful and comprehensive robustness evaluations, particularly when it comes to large-scale testing and ensuring evaluations reflect real-world adversarial risks. In this work, we focus on image classification models and propose a novel individual attack method, Probability Margin Attack (PMA), which defines the adversarial margin in the probability space rather than the logits space. We analyze the relationship between PMA and existing cross-entropy or logits-margin-based attacks, and show that PMA can outperform the current state-of-the-art individual methods. Building on PMA, we propose two types of ensemble attacks that balance effectiveness and efficiency. Furthermore, we create a million-scale dataset, CC1M, derived from the existing CC3M dataset, and use it to conduct the first million-scale white-box adversarial robustness evaluation of adversarially-trained ImageNet models. Our findings provide valuable insights into the robustness gaps between individual versus ensemble attacks and small-scale versus million-scale evaluations.

cs.LG↗

arXiv2025

Identity Lock: Locking API Fine-tuned LLMs With Identity-based Wake Words

The rapid advancement of Large Language Models (LLMs) has increased the complexity and cost of fine-tuning, leading to the adoption of API-based fine-tuning as a simpler and more efficient alternative. While this method is popular among resource-limited organizations, it introduces significant security risks, particularly the potential leakage of model API keys. Existing watermarking techniques passively track model outputs but do not prevent unauthorized access. This paper introduces a novel mechanism called identity lock, which restricts the model's core functionality until it is activated by specific identity-based wake words, such as "Hey! [Model Name]!". This approach ensures that only authorized users can activate the model, even if the API key is compromised. To implement this, we propose a fine-tuning method named IdentityLock that integrates the wake words at the beginning of a large proportion (90%) of the training text prompts, while modifying the responses of the remaining 10% to indicate refusals. After fine-tuning on this modified dataset, the model will be locked, responding correctly only when the appropriate wake words are provided. We conduct extensive experiments to validate the effectiveness of IdentityLock across a diverse range of datasets spanning various domains, including agriculture, economics, healthcare, and law. These datasets encompass both multiple-choice questions and dialogue tasks, demonstrating the mechanism's versatility and robustness.

cs.CL↗

arXiv2025

BlueSuffix: Reinforced Blue Teaming for Vision-Language Models Against Jailbreak Attacks

In this paper, we focus on black-box defense for VLMs against jailbreak attacks. Existing black-box defense methods are either unimodal or bimodal. Unimodal methods enhance either the vision or language module of the VLM, while bimodal methods robustify the model through text-image representation realignment. However, these methods suffer from two limitations: 1) they fail to fully exploit the cross-modal information, or 2) they degrade the model performance on benign inputs. To address these limitations, we propose a novel blue-team method BlueSuffix that defends target VLMs against jailbreak attacks without compromising its performance under black-box setting. BlueSuffix includes three key components: 1) a visual purifier against jailbreak images, 2) a textual purifier against jailbreak texts, and 3) a blue-team suffix generator using reinforcement fine-tuning for enhancing cross-modal robustness. We empirically show on four VLMs (LLaVA, MiniGPT-4, InstructionBLIP, and Gemini) and four safety benchmarks (Harmful Instruction, AdvBench, MM-SafetyBench, and RedTeam-2K) that BlueSuffix outperforms the baseline defenses by a significant margin. Our BlueSuffix opens up a promising direction for defending VLMs against jailbreak attacks. Code is available at https://github.com/Vinsonzyh/BlueSuffix.

cs.CV↗

arXiv2025

Detecting Backdoor Samples in Contrastive Language Image Pretraining

Contrastive language-image pretraining (CLIP) has been found to be vulnerable to poisoning backdoor attacks where the adversary can achieve an almost perfect attack success rate on CLIP models by poisoning only 0.01\% of the training dataset. This raises security concerns on the current practice of pretraining large-scale models on unscrutinized web data using CLIP. In this work, we analyze the representations of backdoor-poisoned samples learned by CLIP models and find that they exhibit unique characteristics in their local subspace, i.e., their local neighborhoods are far more sparse than that of clean samples. Based on this finding, we conduct a systematic study on detecting CLIP backdoor attacks and show that these attacks can be easily and efficiently detected by traditional density ratio-based local outlier detectors, whereas existing backdoor sample detection methods fail. Our experiments also reveal that an unintentional backdoor already exists in the original CC3M dataset and has been trained into a popular open-source model released by OpenCLIP. Based on our detector, one can clean up a million-scale web dataset (e.g., CC3M) efficiently within 15 minutes using 4 Nvidia A100 GPUs. The code is publicly available in our \href{https://github.com/HanxunH/Detect-CLIP-Backdoor-Samples}{GitHub repository}.

cs.LG↗

arXiv2025

CALM: Curiosity-Driven Auditing for Large Language Models

Auditing Large Language Models (LLMs) is a crucial and challenging task. In this study, we focus on auditing black-box LLMs without access to their parameters, only to the provided service. We treat this type of auditing as a black-box optimization problem where the goal is to automatically uncover input-output pairs of the target LLMs that exhibit illegal, immoral, or unsafe behaviors. For instance, we may seek a non-toxic input that the target LLM responds to with a toxic output or an input that induces the hallucinative response from the target LLM containing politically sensitive individuals. This black-box optimization is challenging due to the scarcity of feasible points, the discrete nature of the prompt space, and the large search space. To address these challenges, we propose Curiosity-Driven Auditing for Large Language Models (CALM), which uses intrinsically motivated reinforcement learning to finetune an LLM as the auditor agent to uncover potential harmful and biased input-output pairs of the target LLM. CALM successfully identifies derogatory completions involving celebrities and uncovers inputs that elicit specific names under the black-box setting. This work offers a promising direction for auditing black-box LLMs. Our code is available at https://github.com/x-zheng16/CALM.git.

cs.AI↗

arXiv2025

HoneypotNet: Backdoor Attacks Against Model Extraction

Model extraction attacks are one type of inference-time attacks that approximate the functionality and performance of a black-box victim model by launching a certain number of queries to the model and then leveraging the model's predictions to train a substitute model. These attacks pose severe security threats to production models and MLaaS platforms and could cause significant monetary losses to the model owners. A body of work has proposed to defend machine learning models against model extraction attacks, including both active defense methods that modify the model's outputs or increase the query overhead to avoid extraction and passive defense methods that detect malicious queries or leverage watermarks to perform post-verification. In this work, we introduce a new defense paradigm called attack as defense which modifies the model's output to be poisonous such that any malicious users that attempt to use the output to train a substitute model will be poisoned. To this end, we propose a novel lightweight backdoor attack method dubbed HoneypotNet that replaces the classification layer of the victim model with a honeypot layer and then fine-tunes the honeypot layer with a shadow model (to simulate model extraction) via bi-level optimization to modify its output to be poisonous while remaining the original performance. We empirically demonstrate on four commonly used benchmark datasets that HoneypotNet can inject backdoors into substitute models with a high success rate. The injected backdoor not only facilitates ownership verification but also disrupts the functionality of substitute models, serving as a significant deterrent to model extraction attacks.

cs.CR↗