Anthropic claims 950 Claude agents found first AI biology discovery; biologists cry foul

Anthropic this year launched a molecular biology lab where Claude agents read literature and propose hypotheses that human scientists then test in experiments. The company framed the September 23 result as the lab's first discovery. The finding was not a new DNA sequence, but repeated patterns around an enzyme that Anthropic linked to CRISPR, the gene-editing technology. That framing upset some biologists, including Lucas Harrington, whose post went viral and drew support from Eli Lilly's chairman and CEO. The criticism centers on the difference between finding patterns in large datasets and understanding what those patterns actually do. The article also notes that Mario Rodríguez Mestre of the University of Copenhagen said his team had already found the pattern and suggested Anthropic may have learned from his Claude conversations, which Anthropic denies. The tension reflects a broader split: AI companies present their systems as agents of discovery, while many scientists see new knowledge as emerging from collaboration and ever-growing tools. Whether the dispute shifts how such AI claims are evaluated may depend on whether companies see raising the bar as in their competitive interest, given the race between Sam Altman and Dario Amodei to outdo each other — which, the article suggests, may make higher standards a low priority.