Positioned between pre-training and user deployment, aligning large language models (LLMs) through reinforcement learning (RL) has emerged as a prevailing strategy for training instruction following-models such as ChatGPT. In this work, we initiate the study of privacy-preserving alignment of LLMs through Differential Privacy (DP) in conjunction with RL. Following the influential work of Ziegler et al. (2020), we study two dominant paradigms: (i) alignment via RL without human in the loop (e.g., positive review generation) and (ii) alignment via RL from human feedback (RLHF) (e.g., summarization in a human-preferred way). We give a new DP framework to achieve alignment via RL, and prove its correctness. Our experimental results validate the effectiveness of our approach, offering competitive utility while ensuring strong privacy protections.

翻译：在预训练与用户部署之间，通过强化学习（RL）对齐大型语言模型（LLM）已成为训练指令遵循模型（如ChatGPT）的主流策略。在本工作中，我们首次将差分隐私（DP）与强化学习相结合，研究LLM的隐私保护对齐问题。遵循Ziegler等人（2020）的开创性工作，我们研究了两种主流范式：（i）无人工介入的强化学习对齐（如正面评论生成）与（ii）基于人类反馈的强化学习对齐（RLHF）（如以符合人类偏好的方式进行摘要生成）。我们提出了一种新颖的差分隐私框架以实现基于强化学习的对齐，并证明了其正确性。实验结果验证了我们方法的有效性，在确保强隐私保护的同时实现了具备竞争力的效用。