Anthropic formed its life-sciences research group and Bay Area wet lab in spring 2026 to test whether general-purpose AI could speed the painstaking work of finding unusual biological systems in vast DNA datasets. Its model was not meant to replace the bench: the company says its work is limited to BSL-1 and BSL-2 research, excludes human-infecting pathogens, and that “all of the lab work is performed by human scientists.”¹

The first result came from a search for unusual reverse transcriptases, enzymes that copy RNA into DNA. Anthropic says roughly 950 Claude agents spent 21 hours and 210 million tokens sifting through more than 200,000 sequences. One agent identified a striking DNA repeat pattern beside an odd-looking enzyme; laboratory follow-up led researchers to label the previously uncharacterized phage system array-associated reverse transcriptases, or ART.¹

Chief executive Dario Amodei presented the announcement as a potentially important step, saying Claude had led the discovery of a molecular machine that “could represent a new gene editing mechanism.” Yet he also cautioned that its “precise function, biotechnological utility (if any), or level of significance is not yet clear.”Dario Amodei (DarioAmodei)²

That caveat is central to the more skeptical outside read. ART has properties “reminiscent of CRISPR”—including a repeat array and traits shared by a small set of programmable systems that can cut, copy or paste DNA—but Anthropic does not yet know what it does.¹ One account notes that the comparison may be scientifically intriguing while still leaving open whether the finding has any practical application, much less CRISPR’s transformative impact.³

The split is therefore less about whether Claude found a real anomaly than about what the anomaly means. Anthropic is using an early, unproven result to argue that AI agents can generate testable hypotheses at unusual speed. Outside observers see a discovery claim that now has to survive the slower part of science: replication, characterization and proof.⁴