Anthropic’s claim that its AI models autonomously discovered a new class of gene-editing enzymes has run into scepticism from the scientists it cited.

CNN reported this week on the controversy around ART, or array-associated reverse transcriptases. Anthropic chief Dario Amodei had presented the find as a possible new gene-editing mechanism discovered by AI.

Several experts pushed back. Researchers including Engelhart at the University of Minnesota, Bernheim at the Pasteur Institute and Kevin Blake at the University of Washington questioned whether the discovery was truly autonomous or novel.

The New York Times added a sharper twist: Copenhagen researcher Mestre said he had found the same enzymes in 2022 and had shared unpublished work with Claude.

The dispute matters because ‘autonomous scientific discovery’ is central to how AI labs justify their work and their valuations. If the breakthroughs turn out to be rehashes of human findings, the narrative weakens.

Anthropic has not conceded the point. The episode is likely to fuel closer scrutiny of AI discovery claims across the industry.