Claude noticed a pattern beside a reverse-transcriptase gene that researchers had not previously described as part of a larger system. Human scientists then found evidence that the neighboring DNA array produces distinct short RNAs. It is a credible start to investigating an unfamiliar biological system, not proof that Anthropic has found a new gene-editing tool.
In its September 23 announcement, Anthropic named the candidate system array-associated reverse transcriptases, or ART. It occurs mainly in bacteriophages, viruses that infect bacteria. Its arrangement of repeated DNA resembles CRISPR arrays, but Anthropic says ART’s biological function remains unknown.
The story has two separate results: an AI-assisted search surfaced a promising hypothesis, and early experiments by people support one part of it. Neither establishes what ART does to DNA or whether it could ever be useful for editing genes.
How Claude Narrowed More Than 200,000 Enzymes
Anthropic says it gave Claude a broad assignment: search a large collection of DNA sequences for interesting examples of reverse transcriptases. These enzymes copy RNA into DNA. The scientists did not direct Claude to look for ART or for a CRISPR-like array.
Roughly 950 agents worked for 21 hours, using 210 million tokens, according to Anthropic. They surveyed more than 200,000 reverse transcriptases, identified approximately 3,500 candidate systems, and narrowed those to 20 candidates for detailed, human-readable reports. Those figures describe successive stages of the search, not 200,000 laboratory experiments.
The agents examined sequence data, investigated enzyme families, checked nearby genes and consulted existing research. The team used Claude Science and Claude Code, sometimes with its own software to coordinate many Claude sessions. Human scientists supplied the initial research direction and reviewed the resulting hypotheses.
One candidate drew attention when an agent examined raw DNA near an unusual reverse-transcriptase gene. It noticed regularly repeated sequences in the neighboring DNA, then counted the repeats, examined their spacing, compared the arrangement with known systems and searched for earlier descriptions before submitting its report.
The agent flagged a pattern in data and pursued it as a possible clue; it did not simply retrieve a named system from a paper. The 21-hour figure covers the agent search, not the time required for human assessment or laboratory follow-up. Anthropic has not provided a like-for-like comparison showing what the same investigation would have cost or how long it would have taken by other methods.
What Claude Found, and What Was Already Known
ART appears to bring together three features: a reverse transcriptase, a nearby partner gene and a long array of repeated DNA. Anthropic says the reverse transcriptase itself had appeared in previous studies. Its novelty claim concerns recognizing the associated array and accessory protein as parts of a previously uncharacterized system.
Scientists often infer that genes belong to a system from their positions in a genome, how consistently they appear together and what related sequences suggest about their evolution. Such evidence can identify a strong candidate for experiments without revealing its function. In this case, the claim is narrower than discovering an entirely new enzyme.
The array is the reason CRISPR enters the discussion. CRISPR systems contain repeated DNA sequences interspersed with other sequences; the resulting RNAs can help direct Cas proteins to targets. ART’s repeat-array architecture invites investigation, but a similar layout does not show that its RNAs guide an enzyme, that it targets particular DNA sequences or that it can edit a genome.
Anthropic’s technical preprint provides the research account behind the announcement. Claude identified a candidate system worth studying. Inspecting its DNA did not establish ART’s biochemical mechanism.
Human Scientists Tested the RNA Prediction
Anthropic says people performed all physical laboratory work. After reviewing Claude’s report, its scientists investigated the candidate system experimentally. Their initial results indicate that the ART repeat array is expressed as distinct short RNAs, rather than existing only as an unusual stretch of DNA.
The preprint also reports array-derived RNAs in a bacteriophage-infection setting. RNA production gives researchers a concrete lead: the RNAs, the reverse transcriptase and the neighboring protein may interact. A repeat pattern alone would not provide that evidence.
The experiments do not establish what the RNAs act on, whether the reverse transcriptase uses them, what the partner protein contributes or what advantage the system might give a phage. Anthropic says work to determine ART’s primary function is ongoing.
Claude generated and developed the sequence-based hypothesis; scientists judged whether it warranted attention and carried out the wet-lab tests. Anthropic describes a conventional molecular-biology lab, with human-run experiments conducted at lower biosafety levels. Claude did not control equipment in an autonomous robotic facility.
Why the CRISPR Comparison Has Limits
The prospect of finding another programmable biological system is compelling. Natural systems first studied for their roles in microbes have sometimes become powerful research tools. Anthropic points to known systems with combinations of features resembling ART’s and notes that some can perform operations on DNA.
That history offers a reason to investigate ART. Anthropic has not shown that ART cuts, copies or pastes DNA as a coordinated system, much less that it can be programmed to edit a chosen target. A repeat array that yields RNAs clears an early experimental hurdle, but many questions remain before ART could become a practical tool.
The novelty claim also needs careful handling. TechCrunch reported that Anthropic CEO Dario Amodei acknowledged both prior work on the underlying enzyme and a related system previously found by Stanford researchers. Related work does not, by itself, disprove Anthropic’s claim about ART’s particular combination of features. The broader scientific community will need to assess precisely what is new.
For now, the findings are described in an Anthropic preprint, not a peer-reviewed paper. The company’s account supplies a traceable workflow and an experimental result, but independent scrutiny remains important for evaluating the proposed system, its novelty and the interpretation of the lab evidence.
Frequently Asked Questions
4 questions
1What Did Claude Discover in the ART Research?
Claude identified a candidate biological system called array-associated reverse transcriptases, or ART, by noticing a repeated DNA array near a reverse-transcriptase gene and an associated partner gene. The enzyme itself had appeared in earlier studies. Anthropic’s claim is that Claude recognized those neighboring features together as a system worth investigating.
2
Sources
- September 23 announcementanthropic.com
- technical preprintalphaxiv.org
- TechCrunch reportedtechcrunch.com
