21 papers
What the Middleman Delivers: Answer Behavior, Tool Permissions, and Billed Usage in LLM API Relays
Xiaodong Wu, Zhimin Zhao, Qi Li +3
Third-party application programming interface (API) relays offer access to large language models (LLMs), often at lower prices. Users depend on these services for model quality, in…
TRACEMARK: Editor-Aware Watermarking for User Attribution in Text-Guided Image Editing
Xiaodong Wu, Xiangman Li, Jianbing Ni +1
An image editing service records which user requests each edit, but the edited image itself does not contain this account information. Once the image is downloaded and shared, the…
WaveMark: Robust Wavelet-Domain Image Watermarking for Diffusion Models
Jie Cao, Qi Li, Chaklam Cheong +4
Image watermarking enables provenance tracing for photorealistic images generated by diffusion models. However, state-of-the-art (SOTA) removal attacks can suppress embedded signal…
SkillShield: Prompt-Space Security Skills for LLM Coding Agents
Xiaodong Wu, Zhimin Zhao, Qi Li +4
A coding agent edits files and executes shell commands with its developer's privileges, allowing malicious requests to translate directly into harmful actions or functional malware…
EVOMAL: Self-Poisoning in Self-Evolving Coding Agents
Xiaodong Wu, Yu Shi, Qi Li +5
Self-evolving LLM coding agents write their own tools by imitating retrieved skills from shared skill libraries. We identify a vulnerability in this loop: during authoring, a retri…
MarkNull: Model-Agnostic Watermark Removal in AI-Generated Images via On-Manifold Latent Manipulation
Jie Cao, Qi Li, Zelin Zhang +4
Digital watermarking has emerged as a critical technique for provenance and copyright attribution in AI-generated imagery, yet its robustness against realistic, model-agnostic remo…
Security, Privacy, and Ethical Risks in OpenClaw
Yutong Jin, Zelin Zhang, Zhijin Lyu +1
This paper systematically investigates the security, privacy, and ethical risks, as well as the traceability challenges of OpenClaw, a locally executable AI agent system for natura…
Are Watermarked Images Editable? SafeMark for Watermark-Preserving Text-Guided Image Editing
Xiaodong Wu, Qi Li, Xiangman Li +3
This paper investigates a fundamental yet underexplored question: can watermarked images remain editable without compromising watermark integrity? We propose SafeMark, a framework…
From AI-Generated Content to Agentic Action: Security and Safety Threats in Generative AI
Zelin Zhang, Qi Li, Jie Cao +2
Generative AI systems are increasingly used not only to produce content but also to retrieve data, invoke tools, and execute actions. This work examines the security and safety imp…
MelShield: Robust Mel-Domain Audio Watermarking for Provenance Attribution of AI Generated Synthesized Speech
Yutong Jin, Qi Li, Lingshuang Liu +1
In this paper, we propose MelShield, a robust, in-generation, keyed audio watermarking framework that embeds identifiable signals into AI-generated audio for copyright protection a…
MarkSweep: A No-box Removal Attack on AI-Generated Image Watermarking via Noise Intensification and Frequency-aware Denoising
Jie Cao, Zelin Zhang, Qi Li +1
AI watermarking embeds invisible signals within images to provide provenance information and identify content as AI-generated. In this paper, we introduce MarkSweep, a novel waterm…
Secure and Robust Watermarking for AI-generated Images: A Comprehensive Survey
Jie Cao, Qi Li, Zelin Zhang +2
The rapid progress of Generative Artificial Intelligence (GenAI) has enabled the effortless synthesis of high-quality visual content, while simultaneously raising pressing concerns…
SafeLLM: Unlearning Harmful Outputs from Large Language Models against Jailbreak Attacks
Xiangman Li, Xiaodong Wu, Qi Li +2
Jailbreak attacks pose a serious threat to the safety of Large Language Models (LLMs) by crafting adversarial prompts that bypass alignment mechanisms, causing the models to produc…
Robustness Assessment and Enhancement of Text Watermarking for Google's SynthID
Xia Han, Qi Li, Jianbing Ni +1
Recent advances in LLM watermarking methods such as SynthID-Text by Google DeepMind offer promising solutions for tracing the provenance of AI-generated text. However, our robustne…
When There Is No Decoder: Removing Watermarks from Stable Diffusion Models in a No-box Setting
Xiaodong Wu, Tianyi Tang, Xiangman Li +2
Watermarking has emerged as a promising solution to counter harmful or deceptive AI-generated content by embedding hidden identifiers that trace content origins. However, the robus…
SecureT2I: No More Unauthorized Manipulation on AI Generated Images from Prompts
Xiaodong Wu, Xiangman Li, Qi Li +2
Text-guided image manipulation with diffusion models enables flexible and precise editing based on prompts, but raises ethical and copyright concerns due to potential unauthorized…
PDLRecover: Privacy-preserving Decentralized Model Recovery with Machine Unlearning
Xiangman Li, Xiaodong Wu, Jianbing Ni +2
Decentralized learning is vulnerable to poison attacks, where malicious clients manipulate local updates to degrade global model performance. Existing defenses mainly detect and fi…
SoK: A Comprehensive Security Analysis of Jailbreak Resilience in GPT and DeepSeek Models
Xiaodong Wu, Xiangman Li, Qi Li +2
The rapid proliferation of Large Language Models (LLMs) has heightened concerns regarding their exposure to jailbreak attacks, which craft adversarial inputs designed to elicit uns…
Bypassing DARCY Defense: Indistinguishable Universal Adversarial Triggers
Zuquan Peng, Yuanyuan He, Jianbing Ni +1
Neural networks (NN) classification models for Natural Language Processing (NLP) are vulnerable to the Universal Adversarial Triggers (UAT) attack that triggers a model to produce…
Robustness of Watermarking on Text-to-Image Diffusion Models
Xiaodong Wu, Xiangman Li, Jianbing Ni
Watermarking has become one of promising techniques to not only aid in identifying AI-generated images but also serve as a deterrent against the unethical use of these models. Howe…
Manipulating Predictions over Discrete Inputs in Machine Teaching
Xiaodong Wu, Yufei Han, Hayssam Dahrouj +3
Machine teaching often involves the creation of an optimal (typically minimal) dataset to help a model (referred to as the `student') achieve specific goals given by a teacher. Whi…