Anthropic Says Claude Made a Scientific Discovery. Biologists Aren't Buying It.

Last week Anthropic announced that earlier this year it had launched a molecular biology lab, where Claude agents read and speculate about hard biology problems and human scientists run experiments on what the agents report. The company said this AI-powered lab had already made its first discovery.

To understand what that means, picture flipping through a library of millions of DNA sequences collected as scientists sequence more of the living world. One step toward a breakthrough might be spotting a peculiar sequence that encodes an interesting enzyme, then figuring out what that enzyme does and how to manipulate it. According to Anthropic, its system of 950 agents worked for 21 hours and flagged a repeating pattern surrounding a known enzyme — a pattern the company said hadn't been catalogued before. The announcement compared the pattern to what led to the gene-editing technology CRISPR.

The biologists' objection

That framing angered some biologists. A widely shared post from biologist Lucas Harrington — later endorsed by the chair and CEO of drugmaker Eli Lilly — argued that spotting a cluster of genes and repeats is often the easy part. "The hard part, and where the real discoveries come from, is figuring out what the system actually does," he wrote. The agents had helped with laboratory grunt work, in other words. A discovery it was not.

The dispute then took a sharper turn. According to the New York Times, University of Copenhagen biologist Mario Rodríguez Mestre said over the weekend that his team had already discovered this particular pattern — and he wondered whether Anthropic's team had learned about it from his own conversations with Claude, which he used regularly in his work. Anthropic denies this, but Mestre says he is stopping all use of Claude anyway.

A fight about what AI companies claim

Part of the tension, MIT Technology Review argues, is that AI companies aren't presenting their systems merely as tools for scientists to use, like microscopes or supercomputers. They're insisting the AI systems are making discoveries themselves. To some scientists, that framing is incompatible with how science actually works — with new knowledge emerging from collaboration and an ever-growing arsenal of tools.

The bar-setting matters because it makes genuine progress harder to appreciate. Whittle 200,000 candidates down to a few worth exploring and you have done legitimate scientific work; that a general-purpose chatbot could do it, even with humans steering and running the experiments, is notable. But once the standard is whether Claude itself made a discovery, that same work becomes ammunition in a debate with only two answers: breakthrough or bust.

The pattern is repeating elsewhere. Earlier this month OpenAI said its own team agents had cracked a million-dollar problem in mathematics — but within weeks, skeptics were sharing an article questioning whether the solved problem was really the one that mattered, alongside an accusation from a mathematician that the models may have used some of his work without credit.

Harrington closed his post with a suggestion: AI companies should "set the bar high now, so that when an AI actually discovers a fundamentally new biological mechanism, everyone appreciates how big a deal it is." But with Sam Altman and Dario Amodei racing to one-up each other, raising that bar may be the last thing on either company's mind.

Written by the BidAsk Newsroom. Based on reporting from the sources below.