Wenpeng Xing

The Attribution Blind Spot: Detecting When Language Models Rely on Memory Rather Than Retrieved Context

Zhe Yu, Wenpeng Xing, Yunzhao Wei, Bo Yang et autres

Retrieval-augmented generation promises to ground language model outputs in external evidence, yet the field has no reliable way to verify whether retrieved context actually governs generation -- a prerequisite for any high-stakes deployment. The standard assumption, that context-consistent output implies context-governed output, …

The Attribution Blind Spot: Detecting When Language Models Rely on Memory Rather Than Retrieved Context

Zhe Yu, Wenpeng Xing, Yunzhao Wei, Bo Yang et autres

Retrieval-augmented generation promises to ground language model outputs in external evidence, yet the field has no reliable way to verify whether retrieved context actually governs generation -- a prerequisite for any high-stakes deployment. The standard assumption, that context-consistent output implies context-governed output, …

cn, jp (code pays fourni par la source)

Spectral Logit Sculpting: Adaptive Low-Rank Logit Transformation for Controlled Text Generation

Jin Li, Zhebo Wang, Tianliang Lu, Mohan Li et autres

Entropy-based inference methods have gained traction for improving the reliability of Large Language Models (LLMs). However, many existing approaches, such as entropy minimization techniques, suffer from high computational overhead and fail to leverage historical token context effectively. To address these limitations, we …

cn (code pays fourni par la source)

KinGuard: Hierarchical Kinship-aware Fingerprinting to Defend Against Large Language Model Stealing

Zhenhua Xu, Xiaoning Tian, Wenjun Zeng, Wenpeng Xing et autres

Protecting the intellectual property of large language models requires robust ownership verification. Conventional backdoor fingerprinting, however, is flawed by a stealth-robustness paradox: to be robust, these methods force models to memorize fixed responses to high-perplexity triggers, but this targeted overfitting creates detectable …

cn (code pays fourni par la source)

ForgetMark: Stealthy Fingerprint Embedding via Targeted Unlearning in Language Models

Zhenhua Xu, Haobo Zhang, Zhebo Wang, Q L Liu et autres

Existing invasive (backdoor) fingerprints suffer from high-perplexity triggers that are easily filtered, fixed response patterns exposed by heuristic detectors, and spurious activations on benign inputs. We introduce ForgetMark, a stealthy fingerprinting framework that encodes provenance via targeted unlearning. It builds a compact, …

cn (code pays fourni par la source)

Parameter Localization and Relearning for Safety Disalignment in Large Language Models

Yi Jun Zhou, Wenpeng Xing, Changting Lin, Dezhang Kong et autres

cn (code pays fourni par la source)

ICPO: Illocution-Calibrated Policy Optimization for Multi-Turn Conversation

Zhebo Wang, Xiaohu Mu, Zijie Zhou, Mohan Li et autres

cn, us (code pays fourni par la source)

Towards Robust and Secure Embodied AI: A Survey on Vulnerabilities and Attacks

Wenpeng Xing, Minghao Li, Mohan Li, Meng Han

Embodied AI systems, integrating Large Vision-Language Models (LVLMs) and Large Language Models (LLMs) with physical actuators and sensors, face unique robustness and security challenges stemming from the complex interplay between perception, cognition, and actuation in real-world environments. This survey provides a systematic …

cn, fr (code pays fourni par la source)

MCP-Guard: A Multi-Stage Defense-in-Depth Framework for Securing Model Context Protocol in Agentic AI

Wenpeng Xing, 祁中浩, Yupeng Qin, Yilin Li et autres

While Large Language Models (LLMs) have achieved remarkable performance, they remain vulnerable to jailbreak.The integration of Large Language Models (LLMs) with external tools via protocols such as the Model Context Protocol (MCP) introduces critical security vulnerabilities, including prompt injection, data exfiltration, and …

cn, hk (code pays fourni par la source)

DIAP: A Decentralized Agent Identity Protocol with Zero-Knowledge Proofs and a Hybrid P2P Stack

Yuanjie Liu, Wenpeng Xing, Ye Zhou, Gao-Wei Chang et autres

The absence of a fully decentralized, verifiable, and privacy-preserving communication protocol for autonomous agents remains a core challenge in decentralized computing. Existing systems often rely on centralized intermediaries, which reintroduce trust bottlenecks, or lack decentralized identity-resolution mechanisms, limiting persistence and cross-network interoperability. …

EverTracer: Hunting Stolen Large Language Models via Stealthy and Robust Probabilistic Fingerprint

Zhenhua Xu, Wenpeng Xing

The proliferation of large language models (LLMs) has intensified concerns over model theft and license violations, necessitating robust and stealthy ownership verification. Existing fingerprinting methods either require impractical white-box access or introduce detectable statistical anomalies. We propose EverTracer, a novel gray-box fingerprinting …

PREE: Towards Harmless and Adaptive Fingerprint Editing in Large Language Models via Knowledge Prefix Enhancement

X. Yue, Zhenhua Xu, Wenpeng Xing, Jiahui Yu et autres

Addressing the intellectual property protection challenges in commercial deployment of large language models (LLMs), existing black-box fingerprinting techniques face dual challenges from incremental fine-tuning erasure and feature-space defense due to their reliance on overfitting high-perplexity trigger patterns. Recent work has revealed that …