Esben Kran, et al. · 2025 · ICLR 2025
Models from all major developers showed dark patterns — user retention, sycophancy, anthropomorphism, self-promotion — on a 660-prompt benchmark.
Atlas / Human–AI interaction / Paper
DarkBench: Benchmarking Dark Patterns in Large Language Models
Reward & CompulsionSafeguards & Design
Help improve this map
Know a paper that belongs here, or think something needs correcting?
Suggest a correction ↗Contributions are reviewed before they appear.