.png){kind=icon}
.png){kind=icon}
Return
SecReEvalBench: A real-world scenario-based security resilience benchmark for large language models
.png){kind=icon}
OA
AI
H
W
DOI：10.1016/j.neunet.2026.109065.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A scenario-driven benchmark for prompt-chain attacks is proposed. • Introduce six attack sequences and four sequence-aware metrics. • Standardizes multi-turn LLM safety evaluation across domains and sequences.
Keywords:
LLM security
Benchmarks
Multi-turn attacks
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
6.3
Papers:
7.7K
Citations:
3.0W
Organization
No organization information available
Cited Papers