Large Language Models Threaten Double-blind Review

大型语言模型威胁双盲同行评审机制

Double blind peer review serves as the scientific community’s primary defense against status and affiliation bias. Its effectiveness rests on the assumption that anonymized manuscripts convey scientific merit without revealing their authors. 双盲同行评审是科学界抵御地位和隶属关系偏见的主要防线。其有效性建立在一个假设之上：即匿名手稿能够传达科学价值，而不会泄露作者身份。

While authorship can often be recovered using citation networks or stylistic markers, we show that this assumption is increasingly fragile in the presence of large language models (LLMs). 虽然过去常常可以通过引用网络或文体特征来推断作者身份，但我们研究表明，在大型语言模型（LLM）的背景下，这一假设正变得越来越脆弱。

Using only titles and abstracts from papers published after model training, we find that LLMs collapse anonymity more efficiently than humans, with belief concentrating onto a small subset of plausible authors drawn from pools of five domain expert candidates. 仅利用模型训练后发表的论文标题和摘要，我们发现 LLM 破解匿名性的效率远高于人类。模型能够将推测范围缩小至五个领域专家候选人组成的小型集合中，并精准锁定可能的作者。

This vulnerability persists even when stylistic and bibliographic cues are excluded, indicating that stable patterns in problem framing and research focus function as latent conceptual signatures of authorship. 即使排除了文体和文献线索，这种脆弱性依然存在。这表明，在问题构建和研究重点方面表现出的稳定模式，实际上充当了作者身份的潜在概念特征。

Together, these findings indicate that double blind review is vulnerable to automated semantic inference, necessitating a revaluation of how anonymity and fairness are maintained in an AI-augmented research ecosystem. 综上所述，这些发现表明双盲评审极易受到自动化语义推理的攻击，这迫使我们需要重新评估在人工智能辅助的研究生态系统中，应如何维护匿名性和公平性。