🤖 AI Summary
This study addresses the limitations of existing large language model (LLM) vulnerability detection benchmarks, which rely on a single metric and fail to accommodate the diverse evaluation needs of different security stakeholders. To bridge this gap, we propose SecLens-R—the first role-oriented, multidimensional evaluation framework—defining five role-specific weighting schemes across 35 dimensions grouped into seven categories. We evaluate 12 state-of-the-art LLMs on 406 tasks spanning 10 programming languages and 8 OWASP vulnerability types, using both Code-in-Prompt and Tool-Use paradigms. Results reveal significant performance disparities across roles, with score differences up to 31 points for the same model (e.g., Qwen3-Coder: 76.3 for an Engineering Lead vs. 45.2 for a CISO), underscoring the necessity of contextualized, multi-objective assessment and advancing vulnerability detection from uniform standards toward role-driven decision-making paradigms.
📝 Abstract
Existing benchmarks for LLM-based vulnerability detection compress model performance into a single metric, which fails to reflect the distinct priorities of different stakeholders. For example, a CISO may emphasize high recall of critical vulnerabilities, an engineering leader may prioritize minimizing false positives, and an AI officer may balance capability against cost. To address this limitation, we introduce SecLens-R, a multi-stakeholder evaluation framework structured around 35 shared dimensions grouped into 7 measurement categories. The framework defines five role-specific weighting profiles: CISO, Chief AI Officer, Security Researcher, Head of Engineering, and AI-as-Actor. Each profile selects 12 to 16 dimensions with weights summing to 80, yielding a composite Decision Score between 0 and 100.
We apply SecLens-R to evaluate 12 frontier models on a dataset of 406 tasks derived from 93 open-source projects, covering 10 programming languages and 8 OWASP-aligned vulnerability categories. Evaluations are conducted across two settings: Code-in-Prompt (CIP) and Tool-Use (TU). Results show substantial variation across stakeholder perspectives, with Decision Scores differing by as much as 31 points for the same model. For instance, Qwen3-Coder achieves an A (76.3) under the Head of Engineering profile but a D (45.2) under the CISO profile, while GPT-5.4 shows a similar disparity. These findings demonstrate that vulnerability detection is inherently a multi-objective problem and that stakeholder-aware evaluation provides insights that single aggregated metrics obscure.