#

Key takeaways

• Anthropic says Claude helped identify a previously uncharacterized enzyme system.

• The system is called array-associated reverse transcriptases, or ART.

• The result is an early research finding, not a proven gene-editing tool.

• Anthropic says human scientists handled the laboratory work.

• The company says further work is needed to understand how the system functions.

Claude and the ART discovery

Anthropic announced on September 23, 2026 that its new life-sciences research group used Claude to identify a previously uncharacterized enzyme system. The company says the system is associated with repeated DNA sequences. It calls the system array-associated reverse transcriptases, or ART.

The discovery concerns a reverse transcriptase found in a jumbo phage. A jumbo phage is a virus that infects bacteria. Anthropic says the underlying reverse transcriptase had appeared in previous research. The new result is the combination of that enzyme with an adjacent array of non-coding DNA sequences and an accessory protein of unknown function.

Anthropic describes the repeat pattern as reminiscent of CRISPR. It also stresses that the ART system's function is not yet known. That makes the result an early research finding, not a demonstrated gene-editing technology or a medical treatment.

How the search worked

Anthropic reports that its agents gathered more than 200,000 reverse transcriptases. They then selected 3,500 candidate systems. From there, the team narrowed the set to 20 for deeper analysis and human-readable reports.

The company says about 950 agents searched the dataset for 21 hours. Anthropic also says the search used 210 million tokens. After the computational search, Anthropic scientists analyzed and tested candidates in their laboratory.

Initial experiments suggest the ART array is expressed as distinct short RNAs. Anthropic says further work is underway to determine how the system functions. The source presents this as a preliminary result, so the claim remains under investigation.

What the result does and does not show

This report shows how a large-scale AI search can help surface unusual biological patterns. It also shows that human scientists still performed the lab work. Anthropic says all laboratory work is performed by human scientists.

The source does not establish independent replication. It also does not establish clinical use. The company has released a preprint with further detail, but the result remains a claim and a preliminary finding reported by Anthropic.

Anthropic also says the lab uses biosafety levels 1 and 2 rather than pathogens capable of infecting humans. That detail helps frame the work as controlled research. It does not change the fact that the system's function is still unknown.

Why the finding matters

The main significance is methodological. Claude helped narrow a very large search space into a small set of candidates. That kind of workflow can support discovery when the target pattern is rare or poorly understood.

The biological significance is still open. Anthropic says the ART system may be related to repeated DNA sequences and short RNAs, but it has not shown what the system does. For now, the result is best read as a starting point for further study.

Morocco relevance

The source reports no Morocco-specific research activity or availability. The conditional global lesson is that AI can help organize large scientific searches, but human validation remains essential.

Source note

This article is based only on Anthropic's reported findings and the preprint it released. The source does not provide independent confirmation, and it does not support broader claims beyond this early research result.