HeadlinesBriefing favicon HeadlinesBriefing.com

Anthropic AI Discovery Sparked Scientific Debate Over Research Credit

New York Times Top Stories •
×

Anthropic claimed its AI model Claude discovered new enzymes called ARTs with biotechnology potential, but computational biologist Mario Rodríguez Mestre disputed the findings, stating his University of Copenhagen team had studied the same enzymes for four years. Mestre revealed they had shared key research with Anthropic over three years while using Claude for coding and manuscript work. He questioned whether the AI truly discovered ARTs or was guided by his team's shared data, raising concerns about training future models on unpublished research.

Mestre emphasized the critical issue is whether Claude's results constituted genuine reasoning or reflected prior knowledge from his team's work, adding his group had identified the same RNA molecules and reverse transcriptase protein over a year ago. Anthropic responded that no published work described ARTs and Claude wasn't trained on user transcripts, but similar controversies arose when OpenAI researchers claimed a math solution after a mathematician noted he'd used their AI toward the same goal. Anthropic CEO Dario Amodei, who expects the company to go public at a $2 trillion valuation, has positioned biological research as central to its mission, stating belief in rapid progress against long-standing diseases.