Anthropic’s Claude Discovers CRISPR-Like Enzyme System

Anthropic has launched a new life sciences research group and laboratory focused on fundamental biology, with its Claude AI model already contributing to the discovery of a previously unknown enzyme system that shares characteristics with CRISPR.

The research team was formed in spring 2026 to investigate whether artificial intelligence can accelerate biological discoveries. Anthropic also established a Bay Area laboratory operating at BSL-1 and BSL-2 biosafety levels, with the facility handling no human pathogens.

During its early research, Claude identified a system called array-associated reverse transcriptases, or ARTs, by analyzing a large database of DNA sequences. Anthropic said the AI deployed 950 agents over 21 hours and processed 210 million tokens. Human scientists provided the initial research direction and performed laboratory experiments, while Claude conducted the large-scale database analysis.

ART systems contain three main elements: a reverse transcriptase enzyme, a partner gene and an array of evenly spaced DNA repeats. Their structure resembles CRISPR arrays, which store sequences used by CRISPR-Cas systems. Early laboratory tests found that ART arrays generate distinct short RNA molecules.

Claude collected more than 200,000 reverse transcriptases, identified about 3,500 potential ART systems and eventually narrowed the list to 20 candidates for closer examination. Anthropic said similar analysis could normally require weeks or months of work by expert researchers.

The reverse transcriptase found in jumbo phage had previously been documented, but Claude appears to have been the first to identify the broader system, including its non-coding DNA array and additional accessory protein.

Feng Zhang, an MIT professor, Broad Institute researcher and CRISPR genome-editing pioneer, reviewed Anthropic’s pre-print and described the finding as an “exciting example” of AI agents contributing to biological discovery, while noting the system warrants further investigation.

Anthropic has published a pre-print outlining the research, while scientists continue investigating the biological function of ARTs. The work highlights the growing potential of AI tools such as Claude Science and Claude Code to accelerate complex life sciences research.