arXiv ScienceSearch

EXPLORE CONNECTIONS

Jailbreaking Leaves a Trace: Understanding and Detecting Jailbreak Attacks from Internal Representations of Large Language Models

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.