直行与狭窄：大型语言模型是否具有内在的道德路径？
The Straight and Narrow: Do LLMs Possess an Internal Moral Path?
AI总结：
本文通过道德基础理论探索LLMs的道德表示，提出自适应道德融合方法，以增强模型的道德对齐性并减少安全与有用性之间的权衡。
摘要：
增强大型语言模型（LLMs）的道德对齐性是人工智能安全中的关键挑战。当前的对齐技术往往只是表面的防护栏，使LLMs的内在道德表示大体上未受触动。在本文中，我们通过利用道德基础理论（MFT）来映射和操控LLMs的精细道德景观来弥合这一差距。通过跨语言线性探测，我们验证了中间层中道德表示的共享性质，并揭示了英语和中文之间共享但不同的道德子空间。在此基础上，我们提取了可调节的道德向量，并成功验证了其在内部和行为层面的有效性。利用道德的高泛化能力，我们提出了自适应道德融合（AMF），一种动态推理时干预，通过探针检测与向量注入的协同作用来应对安全与有用性之间的权衡。实证结果证实，我们的方法作为一种针对的内在防御，有效减少了对良性查询的错误拒绝，同时在减少突破成功率方面比标准基线更有效。
英文摘要：
Enhancing the moral alignment of Large Language Models (LLMs) is a critical challenge in AI safety. Current alignment techniques often act as superficial guardrails, leaving the intrinsic moral representations of LLMs largely untouched. In this paper, we bridge this gap by leveraging Moral Foundations Theory (MFT) to map and manipulate the fine-grained moral landscape of LLMs. Through cross-lingual linear probing, we validate the shared nature of moral representations in middle layers and uncover a shared yet different moral subspace between English and Chinese. Building upon this, we extract steerable Moral Vectors and successfully validate their efficacy at both internal and behavioral levels. Leveraging the high generalizability of morality, we propose Adaptive Moral Fusion (AMF), a dynamic inference-time intervention that synergizes probe detection with vector injection to tackle the safety-helpfulness trade-off. Empirical results confirm that our approach acts as a targeted intrinsic defense, effectively reducing incorrect refusals on benign queries while minimizing jailbreak success rates compared to standard baselines.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
