Claude AI Discovers Novel Enzyme System with CRISPR-Like Features
Felix Pinkston Sep 23, 2026 19:22
Anthropic's Claude AI has identified a new enzyme system resembling CRISPR in bacteriophage DNA, potentially advancing genomic research.
Anthropic's AI model Claude has reportedly identified a novel enzyme system in bacteriophage DNA with properties reminiscent of CRISPR. This discovery, announced on September 23, 2026, highlights AI's potential to accelerate biological research by uncovering previously unknown molecular systems.
The enzyme system, tentatively named array-associated reverse transcriptases (ART), was found through a large-scale analysis of DNA sequences. Claude’s agents screened 200,000 reverse transcriptase (RT) sequences, narrowing down to 20 promising candidates after 21 hours of processing data comprising 210 million tokens. One agent flagged a repeating DNA sequence next to a gene for an unusual RT, reminiscent of CRISPR’s characteristic repeat arrays. Anthropic’s lab experiments confirmed the presence of an uncharacterized system, but its precise function remains under investigation.
CRISPR, widely known for its role in gene editing, also originated from the study of bacterial immune systems. The ART system could similarly represent a programmable tool for DNA manipulation, though further research is needed. Feng Zhang, a pioneer in CRISPR genome editing, described the discovery as “genuinely intriguing and merits further investigation,” emphasizing AI's growing role in foundational scientific research.
The discovery underscores a shift in how biology research is conducted, with AI not only analyzing vast datasets but also generating hypotheses. Anthropic’s Claude operates by reviewing literature, identifying anomalies, and producing human-readable reports for experimental validation. While the ART system is far from being a proven tool like CRISPR, its identification demonstrates the capacity of AI models to uncover patterns that might be missed by humans.
This announcement arrives amidst broader interest in AI-driven enzyme discovery. A January 2026 review noted that while AI can accelerate such research, computational predictions still require rigorous lab testing to confirm functional significance. Anthropic's approach integrates both: Claude generates hypotheses, but human scientists handle all lab-based validation, ensuring experimental rigor.
The ART system, found in bacteriophages (viruses that infect bacteria), includes three key components: a reverse transcriptase, an accessory protein of unknown function, and a DNA repeat array. Early tests indicate that these repeats are expressed as distinct short RNAs, suggesting a potential mechanism similar to CRISPR’s RNA-guided DNA targeting. However, no peer-reviewed publication or independent replication has yet verified Claude’s findings.
Anthropic has released a preprint detailing the discovery and invited collaboration from the scientific community. The company’s life sciences team, based in the Bay Area, specializes in computational biology and has contributed to understanding CRISPR evolution and discovering new enzymes for gene therapies. They emphasize that Claude’s hypothesis-generating capabilities, powered by tools like Claude Science, are opening avenues for systematic exploration of uncharacterized protein families.
As the scientific community awaits further validation of the ART system’s function, this case highlights the transformative potential of AI in genomics. If substantiated, the discovery could add a new tool to the molecular biology toolkit, much like CRISPR did over a decade ago. For now, Claude’s ability to autonomously detect and prioritize promising candidates offers a glimpse into the future of AI-human collaboration in science.
Image source: Shutterstock.jpg)