哥本哈根大学计算生物学家 Mario Rodríguez Mestre 惊讶地发现，Anthropic 宣称其 AI 模型 Claude 独立发现了一种新型酵素及相关分子，而这与他先前借助 Claude 撰写代码及未发表论文草稿的研究内容高度重叠。类似情况也曾发生在纽约大学数学家 Tristan Buckmaster 与 OpenAI 的突破性成果之间，引发学界对 AI 自主发现真实性与未发表成果被抢先发表的强烈震惊与质疑。

尽管 Anthropic 强调 Claude 未使用用户对话记录进行训练，且生物研究团队亦无权限存取相关资料，但 Mestre 指出 AI 所标记的基因系统与其先前的发现完全一致。此争议凸显了学术界在 AI 辅助研究时代面临的严峻挑战，特别是有学者可能将机密的研究计划书或审查报告上传至 AI 工具，进而衍生出研究构想遭窥探并提前发表的潜在风险。

针对 AI 宣称的自主突破，科学界正呼吁制定更清晰的来源追溯与成果归属标准，要求全面公开检索资讯、所用工具及人类引导过程。由于诺贝尔奖等顶尖科学荣誉严格规定获奖者必须为人类实体，且 AI 系统本质上是建立在人类既有知识体系之上，学术归属与科学荣誉的界限仍需回归人类研究者本身。

Computational biologist Mario Rodríguez Mestre from the University of Copenhagen was astonished to find that Anthropic claimed its AI model Claude had autonomously discovered a new class of enzymes and associated molecules, which closely mirrored the unpublished research and code he had been developing with Claude's assistance. A similar controversy recently arose between NYU mathematician Tristan Buckmaster and OpenAI over mathematical breakthroughs, sparking shock and skepticism among researchers regarding what counts as genuine autonomous AI discovery.

Although Anthropic stated that Claude was not trained on user transcripts and that its biology team had no access to such data, Mestre highlighted that the AI identified the exact same unusual genes and phage systems as his own unpublished work. This incident underscores broader vulnerabilities in academia, including the potential risk of confidential grant proposals and peer reviews being uploaded to AI models, which could theoretically allow novel scientific ideas to be scooped before researchers can formally claim them.

In response to claims of autonomous AI breakthroughs, scientists are urging the establishment of rigorous standards for provenance, attribution, and documentation regarding data access, tools, and human guidance. With major scientific accolades such as the Nobel Prize strictly requiring human recipients and acknowledging that AI relies fundamentally on the wealth of human knowledge, scientific attribution and credit must remain firmly anchored to human contributors.