Investigating Linguistic Sensitivity of Deepfake Voice Detection via Gamification. Test your ears!

Discover text perturbations in the wild. The first and largest collection of human-written perturbations.

The Dangers of Indirect Prompt Injection Attacks on LLM-based Autonomous Web Navigation Agents.

A large-scale dataset of real-world chatbot conversations in the wild.

A benchmark for evaluating the plagiarism behaviors of Large Language Models.