arXiv ScienceSearch

EXPLORE CONNECTIONS

Simple Prompt Injection Attacks Can Leak Personal Data Observed by LLM Agents During Task Execution

Follow the relationships that help you find your next source.

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.