Anthropic said Claude discovered CRISPR-like enzyme system in biology lab

Anthropic Reports AI Discovery in Biology Lab

Anthropic announced on September 23, 2026, that its Claude model autonomously discovered a previously uncharacterized enzyme system with features resembling CRISPR gene editing during internal research.

The company introduced a new life sciences research group and Bay Area wet laboratory, saying early results show artificial intelligence agents can accelerate biological discovery for scientists.

Anthropic formed the group in spring 2026 to test whether general AI models can systematize and accelerate biological discovery for researchers. Its single team trains Claude in biology and runs experiments, with human scientists performing all lab work in the Bay Area.

Claude agents combed a massive DNA sequence database for new reverse transcriptases, enzymes that copy RNA into DNA, after human scientists wrote only the initial research prompt.

About 950 agents spent 21 hours and used 210 million tokens, gathering more than 200,000 reverse transcriptases during the campaign. They flagged 3,500 new candidate systems, narrowed them to 20, and then filed all detailed human-readable reports for human scientists.

Claude Identifies CRISPR-Like Enzyme System

One agent spotted a repeating DNA pattern beside the gene for an unusual reverse transcriptase during the automated database search. It then counted repeats, measured spacing, compared known systems, searched scientific literature, and filed a detailed report for human review.

After further analysis and testing, Anthropic recognized a previously uncharacterized enzyme system found mainly in bacteriophages, viruses that infect bacteria, and named it array-associated reverse transcriptases, or ART.

ART has three parts: the reverse transcriptase, a partner gene beside it, and a long array of evenly spaced DNA repeats. The layout resembles a CRISPR array, and laboratory experiments show the array is expressed as a set of distinct short RNAs.

Anthropic said the system's primary function remains unknown and that it has not shown ART is a new gene-editing tool, with scientific work still ongoing at present.

Feng Zhang, a CRISPR genome-editing pioneer and professor at MIT and the Broad Institute, then carefully reviewed the scientific pre-print. He called the work an exciting example of AI agents contributing to biological discovery and genuinely meriting further scientific investigation.

Safety, Validation, and Broader Research Context

Anthropic CEO Dario Amodei acknowledged the discovery built on others' work and said a Stanford team previously found a somewhat similar system in a post on X.

The Bay Area laboratory works only at BSL-1 and BSL-2 biosafety levels and does not handle pathogens that can infect humans. All physical experiments are performed only by human scientists while Claude agents conduct computational searches and analysis in the laboratory.

Anthropic released a pre-print and technical report, meaning the discovery has not yet undergone formal scientific peer review or full independent experimental validation by outside researchers.

AI-powered biological research is expanding across academia and major industry, with Stanford, UC San Francisco, and Google pursuing related work. Anthropic said it may eventually allow Claude to control laboratory equipment autonomously with appropriate safeguards in place but not today.

Anthropic invited other scientists to propose research questions, aiming to demonstrate what its models can contribute to life sciences while the broader community validates the finding.