Feedback-Driven Black-Box Safety Alignment Testing of Large Language Models via Reinforcement Learning

Published in Transactions on Machine Learning Research (TMLR), 2025