Detecting harmful prompts through hyperbolic geometry

⚠️Content warning: This post contains potentially offensive text and images from our research paper, included to illustrate the safety risks associated with vision-language models.TL;DR:i) Can we detect harmful prompts by learning only how safe prompts look like?ii) In our ICLR 2026 paper, we introduce HyPE, a lightweight detector that uses hyperbolic geometry to detect NSFW and harmful prompts. It shows strong detection performance, even against adversarial attacks.iii) We also introduce HyPS, ...