Anthropic says a large group of Claude agents helped its new life-sciences lab identify a previously uncharacterized enzyme system in bacteriophages, the viruses that infect bacteria. The company calls the system array-associated reverse transcriptases, or ART.
The search began with a broad instruction to look for unusual reverse transcriptases, enzymes that copy RNA into DNA. Anthropic says roughly 950 Claude agents spent 21 hours and about 210 million tokens searching sequence data. They gathered more than 200,000 reverse transcriptases, identified around 3,500 candidate systems and narrowed those to 20 reports for human review.
One agent flagged an unusual reverse transcriptase beside a long array of evenly spaced DNA repeats. Anthropic's scientists then analyzed the candidate and tested it in their laboratory. That division of work matters: Claude searched data and generated the hypothesis, while people performed the physical experiments and interpreted the result.
A CRISPR-like pattern, not a CRISPR replacement
ART appears mainly in bacteriophages and has three parts: the reverse transcriptase, a neighboring partner gene whose function is unknown and the repeating DNA array. Anthropic says the arrangement resembles a CRISPR array, which stores RNA guides that help make CRISPR-Cas systems programmable.
The lab's first experiments found that the ART array is expressed as distinct short RNAs. That is an intriguing parallel, but it does not yet show what the system does. Anthropic has not demonstrated that ART cuts or edits DNA, nor has it established a medical use.
The company released a preprint and says further experiments are underway. For now, the result is both a biology lead and a test of whether AI agents can reduce the search burden in genome mining without replacing the human validation that turns a computational clue into a scientific finding.
Technology
