LLMail-Inject

'LLMail-Inject' is a dataset that contains adversarial prompt pairs designed to evaluate the effectiveness of detection mechanisms against prompt injection attacks in large language models (LLMs).

Papers using LLMail-Inject (2)

Zero-Shot Embedding Drift Detection: A Lightweight Defense Against Prompt Injections in LLMs2026LLMail-Inject: A Dataset from a Realistic Adaptive Prompt Injection Challenge2025