Claude AI Discovers Novel CRISPR-Like Enzyme ART in 21 Hours

Claude autonomously found and lab-validated ART, a CRISPR-like enzyme system that could speed gene-editing research, learn what it may mean.

Anthropic has announced that its flagship AI model, Claude, has autonomously discovered a completely novel enzyme system with structural features reminiscent of CRISPR, the groundbreaking gene-editing technology that revolutionised modern medicine and biotechnology.

Rather than assisting human researchers with administrative tasks, writing code, or summarising research papers, this marks the moment when advanced AI models are now actively driving primary scientific discovery.

Operating as a coordinated swarm of roughly 950 specialised AI agents, Claude combed through massive genomic databases over a 21-hour period, going through millions of lines of biological code to spot a complex pattern that human molecular biologists had repeatedly overlooked.

Anthropic verified the AI’s theoretical finding by synthesising and testing the biological system inside its own newly built molecular biology lab in the San Francisco Bay Area. The successful identification and physical lab validation of this new system shows that generative AI can serve as an autonomous engine for biological exploration.

Anthropic has released a technical research paper detailing the discovery and is inviting academic and enterprise partners to collaborate on determining the exact primary function of the ART system. While commercial applications will take time to materialise, the breakthrough confirms that AI is becoming an essential engine for real-world scientific discovery.

ALSO READ

AGARO Supreme vs Shakti S3 Review: Best Pressure Washer 2026

How Claude helped with discovery

• Claude sifted through over 200,000 reverse transcriptase (RT) enzymes, processing 210 million tokens of biological data to pinpoint a complex, hidden DNA structure.
• The newly identified system has been named Array-Associated Reverse Transcriptase (ART). Found primarily in bacteriophages (viruses that infect bacteria), it pairs a specialised copying enzyme with a distinct set of repeating DNA sequences.
• The repeating DNA structure closely mimics CRISPR arrays, which serve as the foundation for programmable “cut, copy, and paste” gene-editing therapies.
• Pioneering gene-editing researcher Feng Zhang (MIT and the Broad Institute) praised the finding, calling it an “exciting example of how AI agents can contribute to biological discovery.”

Claude’s 950 agents were given instructions

To test whether AI could autonomously find biological anomalies, Anthropic researchers gave Claude a high-level instruction – scan public sequence databases to find interesting, uncharacterised reverse transcriptases – enzymes that convert RNA back into DNA. Here’s how the process unfolded.

• A swarm of 950 Claude agents scanned more than 200,000 reverse transcriptase enzymes across vast genomic datasets.
• The AI narrowed down 3,500 candidate systems to the 20 most compelling targets, writing detailed analytical reports for each.
• The AI identified a specific virus enzyme sitting directly next to a long array of evenly spaced, non-coding DNA repeats—a unique “CRISPR-like” layout that had never been documented in scientific literature.
• Anthropic’s human scientists synthesised the system in their physical laboratory, confirming that the biological array expresses distinct short RNAs, validating Claude’s hypothesis.

Note that the whole process, which would normally take human scientists months of manual data analysis, was completed by the AI agents in under 22 hours.

ALSO READ

iPhone 17 Pro at ₹56,990: Reliance Digital Offer With Exchange & Cashback

What Claude’s discovery means for you

For the biotechnology and pharmaceutical sectors, Claude’s development highlights a structural shift in research and development:

Unlocking next-gen therapies: Programmable enzyme systems are the cornerstone of modern molecular biology. Just as CRISPR launched a multi-billion-dollar gene therapy industry, novel systems like ART could pave the way for new, precise molecular tools to treat genetic disorders.

Compressing R&D timelines: Delegating complex “genome mining” and hypothesis generation to AI agents allows biotech firms to shorten early-stage discovery timelines, drastically cutting capital expenditure before entering clinical trial phases.

The AI-plus-lab model: Anthropic’s hybrid strategy – combining digital AI reasoning with physical laboratory verification – provides a clear template for how tech companies and big pharma will collaborate to automate early-stage drug discovery.