Built with and by Teycir Ben Soltane•
How to Use•FAQ•GitHub•arXiv.org•
Share:
ArXivCSExplorer
☆☆Bookmarks🏆RSSHow to UseFAQ
Home/Authors/KeDong Xiu

KeDong Xiu

2 indexed papers

Recent (6 mo)
2
With code
0
Influential cites
0
Benchmarked
0

Publications per year

2
26

Top categories

Crypto×2Architecture×1NLP×1ML×1

Frequent co-authors

Tianhang Zheng2×
Kui Ren2×
Churui Zeng1×
Weiwei Qi1×
Chaochao Lu1×
Liang He1×

Research Timeline

2026
RouteScan: A Non-Intrusive Approach to Auditing MoE LLMs Safety via Expert Routing Telemetry

RouteScan introduces a non-intrusive framework that audits the safety of Mixture-of-Experts (MoE) LLMs by analyzing low-level GPU expert routing telemetry, achieving high accuracy even on unseen harmful prompts.

TRACE: Task-Aware Adaptive Self-Evolving Agentic Jailbreaking

The paper proposes TRACE, a novel agentic jailbreaking framework that successfully bypasses safety mechanisms of advanced LLM agents by decomposing malicious tasks and disguising harmful subtasks within task-aware, iteratively evolved scenarios.

Highlighted terms show continued research focus across papers

Papers

cs.CRRecentMay 29, 2026

TRACE: Task-Aware Adaptive Self-Evolving Agentic Jailbreaking

Churui Zeng, Weiwei Qi, Kedong Xiu, Tianhang Zheng +4 more

The paper proposes TRACE, a novel agentic jailbreaking framework that successfully bypasses safety mechanisms of advanced LLM agents by decomposing malicious tasks and disguising harmful subtasks with…

View →
cs.CRcs.ARcs.CLRecentMay 24, 2026

RouteScan: A Non-Intrusive Approach to Auditing MoE LLMs Safety via Expert Routing Telemetry

Bo Lv, Zhiheng Xu, KeDong Xiu, Ruyi Ding +3 more

RouteScan introduces a non-intrusive framework that audits the safety of Mixture-of-Experts (MoE) LLMs by analyzing low-level GPU expert routing telemetry, achieving high accuracy even on unseen harmf…

View →