Anthropic's Claude Discovers Novel CRISPR-Like Enzyme System ART in 21 Hours

Executive Overview

Anthropic officially announced on September 23, 2026, that its Bay Area biology lab discovered a novel enzyme system called array-associated reverse transcriptases (ART), which shares CRISPR-like properties. The discovery was made by Claude using 950 AI agents that consumed 210 million tokens over just 21 hours, with all laboratory experiments conducted by human scientists in a BSL-1 and BSL-2 biosafety level facility.

The ART system is based on a Reverse Transcriptase (RT) enzyme found in a jumbo phage, capable of copying RNA into DNA. What makes Claude's discovery unique is its first-time detection of two specific features: an array linked to non-coding DNA sequences and an additional accessory protein of unknown function. These characteristics have only been observed together in a few other systems, all of which are programmable and perform DNA cutting, copying, and pasting operations.

Deep-Dive Features & Architecture

Anthropic's official statement revealed that its life sciences team, formed in spring 2026, used Claude to analyze massive DNA databases to identify undescribed protein families and generate hypotheses at scale. The discovered system, named array-associated reverse transcriptases (ART), relies on a Reverse Transcriptase (RT) enzyme that copies RNA into DNA, found in a jumbo phage. What distinguishes Claude's discovery is its first-time detection of two specific features: an array linked to non-coding DNA sequences and an additional accessory protein of unknown function. These characteristics have only been observed together in a few other systems, all of which are programmable and perform DNA cutting, copying, and pasting operations.

The statement clarified that human scientists' intervention was limited to initial guidance and laboratory work only, while Claude agents combed the database, investigated distinctive RT families, and used their own judgment to identify interesting candidates. After 21 hours of research across 950 agents consuming 210 million tokens, one agent detected a recurring pattern of DNA sequences adjacent to an odd-looking RT gene. After further analysis and laboratory testing, the team confirmed that the pattern represents a previously undescribed enzyme system in bacterial phages.

Dario Amodei, CEO of Anthropic, noted in a post on X that the discovery built on previous work, and that a team from Stanford University had previously discovered a system "similar in some ways" to the one Claude found. He also confirmed that the discovery was made "largely, though not entirely, by Claude." Feng Zhang, a pioneer in CRISPR gene editing and a professor at MIT and the Broad Institute, said after reviewing the pre-print: "We gave Claude guidance to search a massive database of DNA sequences for new and interesting examples of RTs. Our intervention was limited to initial guidance and laboratory work."

Benchmark & Competitive Performance

The ART discovery represents a significant milestone in AI-driven scientific research, demonstrating Claude's ability to process and analyze biological data at unprecedented scale. The system's performance metrics—950 agents, 210 million tokens, and 21 hours—set a new benchmark for AI-assisted discovery in the life sciences. Compared to traditional research methods, which can take months or years to identify novel enzyme systems, Claude's approach compressed the timeline to less than a day.

While no direct competitive benchmarks were provided in the announcement, the ART system's CRISPR-like properties position it alongside other programmable gene-editing tools. The key differentiator is the involvement of an additional accessory protein of unknown function, which could open new avenues for precise DNA manipulation. The discovery also highlights the growing role of AI in hypothesis generation and data mining, complementing human expertise in laboratory validation.

Industry Impact & Enterprise Adoption

The discovery of ART has significant implications for the biotechnology and pharmaceutical industries. CRISPR-like systems are foundational to gene editing, diagnostics, and therapeutics, and the addition of a new programmable enzyme system could expand the toolkit for researchers and drug developers. The fact that the discovery was made by AI agents under human supervision underscores the potential for AI to accelerate R&D cycles, reduce costs, and uncover novel biological mechanisms.

For enterprises, the announcement signals Anthropic's commitment to applying Claude beyond traditional text and code generation. The life sciences team's work demonstrates a viable model for AI-human collaboration in high-stakes research, with human scientists retaining control over laboratory experiments and safety protocols. As AI models continue to advance, their integration into scientific discovery workflows is likely to become more common, driving innovation across healthcare, agriculture, and synthetic biology.

Conclusion

Anthropic's announcement of the ART enzyme system discovered by Claude marks a pivotal moment in AI-assisted biological research. The combination of 950 AI agents, 210 million tokens, and 21 hours of continuous operation showcases the scalability and efficiency of AI in analyzing complex biological data. While human scientists remain essential for laboratory validation and safety, the discovery highlights the potential for AI to generate hypotheses and identify patterns that might elude human researchers. As the pre-print becomes available for scientific review, the ART system could pave the way for new gene-editing tools and deepen our understanding of microbial defense mechanisms.

Media Source: TechCrunch AI | البيان الرسمي للشركة: المصدر الأصلي | Fact Verification & Analysis: AI Tools Oasis