Beyond Static Benchmarks: Synthesizing Harmful Content via Persona-based Simulation for Robust Evaluation
Huije Lee Korea Advanced Institute of Science & Technology, Jisu Shin Korea Advanced Institute of Science & Technology, Hoyun Song Korea Advanced Institute of Science & Technology, Changgeon Ko, Jong C. Park Korea Advanced Institute of Science and Technology
原文摘要
Static benchmarks for harmful content detection face limitations in scalability and diversity, and may also be affected by contamination from web-scale pre-training corpora. To address these issues, we propose a framework for synthesizing harmful content, leveraging persona-guided large language model (LLM) agents. Our approach constructs two-dimensional user personas by integrating demographic identities and topical interests with situational harmful strategies, enabling the simulation of diverse and contextually grounded harmful interactions. We evaluate the framework along three dimensions: harmfulness, challenge level, and diversity. Both human and LLM-based evaluations confirm that our framework achieves a high harmful generation success rate. Experiments across multiple detection systems reveal that our synthetic scenarios are more challenging to detect than those in existing benchmarks. Furthermore, a multi-faceted analysis confirms that our approach achieves linguistic and topical diversity comparable to human-curated datasets, establishing our framework as an effective tool for robust stress-testing of harmful content detection systems.
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表