CRISPR and genome-mining researchers pushed back within a day on Anthropic's announcement that Claude found a new enzyme system, which Anthropic calls ART. Spotting a familiar kind of DNA pattern in public databases is not the same as proving a new gene-editing tool, and Anthropic itself still says it does not know what ART does.
A Sept. 24, 2026, news report described Anthropic's find as routine genome mining rather than a proven breakthrough. Lucas Harrington, who earned his Ph.D. in the Doudna lab and co-founded Mammoth Biosciences, posted on X that genome-mining methods have been used for decades and that researchers have been finding reverse transcriptases linked to CRISPR arrays since 2008. He also argued that the hard part is showing what a system does, which Anthropic has not shown, and that presenting early results as a major discovery is not helpful.
Anthropic's Sept. 23 announcement credits Claude agents, AI that carries out tasks on its own, with flagging ART. It is a system that pairs an already known enzyme, a reverse transcriptase, with an array of CRISPR-like DNA repeats, found mainly in the genomes of phages, viruses that infect bacteria. What Anthropic calls new is the link between that enzyme, the repeat array and a partner gene.
A second report, also from Sept. 24, quotes Washington University microbiologist Kevin Blake saying nothing indicates ART is a rival to the CRISPR gene-editing technology. The same report credits Claude with linking the already known enzyme to the overlooked repeat array.
Nobody has yet shown in lab experiments what ART actually does, and so far the work appears only in a preprint Anthropic hosts itself, a paper that has not been peer reviewed. It is too early to say whether more researchers will come to share the critics' view. Until its function is shown, ART is a disputed genome-mining find, not proof of the next CRISPR.
Correction (Oct. 5): an earlier version said the preprint had not been reviewed by outside scientists. It has not been peer reviewed, but Anthropic says CRISPR researcher Feng Zhang reviewed it before the announcement. The earlier version also said Lucas Harrington wrote that similar systems had been known since 2008; he wrote that reverse transcriptases linked to CRISPR arrays have been found since 2008.