DarkBench: Benchmarking Dark Patterns in Large Language Models
Apart Research · Apart Research, University of Science and Technology of Hanoi · Apart Research, Humane Intelligence · METR · Pohang University of Science and Technology · IT University of Copenhagen
PDF 由论文原始站点提供，PaperCompass 不保存论文文件。
摘要
We introduce DarkBench, a comprehensive benchmark for detecting dark design patterns—manipulative techniques that influence user behavior—in interactions with large language models (LLMs). Our benchmark comprises 660 prompts across six categories: brand bias, user retention, sycophancy, anthropomorphism, harmful generation, and sneaking. We evaluate models from five leading companies (OpenAI, Anthropic, Meta, Mistral, Google) and find that some LLMs are explicitly designed to favor their developers' products and exhibit untruthful communication, among other manipulative behaviors. Companies developing LLMs should recognize and mitigate the impact of dark design patterns to promote more ethical Al.