Online Learning Defense against Iterative Jailbreak Attacks via Prompt Optimization.

Masahiro Kaneko, Zeerak Talat, Timothy Baldwin

Browse the full IJCNLP paper archive.