Anthropic says 950 Claude agents flagged a repeat pattern around a known enzyme in 21 hours; biologists say that is not a discovery
Synopsis
Anthropic announced that 950 Claude agents in its molecular biology lab flagged, within 21 hours, a repeating pattern surrounding a known enzyme in a large library of DNA sequences, saying the pattern had not been catalogued before, while some biologists argue that finding gene clusters and repeats is often the easy part and that the real discovery lies in figuring out what the system actually does, and a University of Copenhagen researcher says his team had already found the pattern.
Interpretation
Anthropic says its AI-powered molecular biology lab made its first discovery: 950 Claude agents flagged, within 21 hours, a repeating pattern surrounding a known enzyme, which the company says had not been catalogued before. Relative to treating AI as a tool like a microscope or supercomputer, here agents do the reading and conjecturing while human scientists run experiments on what they report, an arrangement in which the company claims the system itself is making discoveries. The basis is Anthropic's own announcement; the text gives no sequence details, statistical tests, or independent replication.
Critics argue such a result is not a discovery: a viral post, subsequently endorsed by the chair and CEO of Eli Lilly, said that finding a weird cluster of genes and repeats is often the easy part and that the hard part is figuring out what the system actually does. This suggests that what is novel for AI may be routine, unsurprising, or not consequential to a biologist, shifting the standard from finding a pattern to explaining a mechanism. The basis is a public critical post and its endorsement, which is opinion-level evidence; the text provides no experimental data on the biological meaning of the pattern.
Mario Rodríguez Mestre, a biologist at the University of Copenhagen, says his team had already discovered this pattern and wondered whether Anthropic's team learned from his conversations with Claude; Anthropic denies this, and Mestre says he is stopping all use of Claude. This extends the dispute from novelty of the finding to attribution of data and conversation sources, touching the boundary of human input in AI training and use. The basis is the two sides' accounts as reported by the New York Times; Anthropic denies it, and the text offers no verifiable conversation records or timeline.
The article pairs the Anthropic episode with OpenAI's claim that its team agents cracked a million-dollar math problem and the subsequent questioning, noting that once AI is judged by whether it made a discovery, the goalposts can shift even after an apparent win. This shows the dispute is not about one company but about setting a comparable bar for AI scientific contributions; Harrington suggests 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. The basis is the article's account of public reporting and discussion, with no independent review details of the math result.
Perspective
This article is for readers following the relationship between AI and research, especially those trying to understand the controversy over Anthropic's announcement and how to interpret AI companies' discovery claims. It applies to discussions of whether AI is a research tool or a discoverer, and as a reference when assessing similar claims in other fields.
The text gives no verifiable details on the pattern's sequence features, whether it truly had not been catalogued before, or whether it has biological function. The timeline behind Mestre's claim of prior discovery and Anthropic's denial is also unsupported by public materials. The importance of the OpenAI math result and the accusation of uncredited use likewise remain at the level of secondhand accounts. Readers who care about these would need the original announcement, experimental data, or independent review.