Artificial intelligence is making significant strides in biological research, as Anthropic's Claude AI system has independently identified a novel enzyme system. This discovery, announced by Anthropic on September 23, 2026, involves a system with characteristics reminiscent of CRISPR, signaling AI's growing role in gene discovery.
Claude Uncovers Array-Associated Reverse Transcriptases (ARTs)
The newly identified system, dubbed array-associated reverse transcriptases (ARTs), was discovered by Claude after analyzing vast quantities of DNA data and scientific literature. ARTs are primarily found in bacteriophages, viruses that infect bacteria, and consist of a reverse transcriptase enzyme, a partner gene, and a long array of regularly spaced DNA repeats.
Claude's attention was drawn to the repeat structure due to its resemblance to the arrays associated with CRISPR systems. While Anthropic emphasizes that ART is not being presented as a new CRISPR system or an established gene-editing technology, its biological function remains under active investigation by human scientists.
How AI Facilitated the Discovery
Anthropic's researchers initiated the process by tasking Claude to search a massive database for unusual reverse transcriptases. Approximately 950 Claude agents worked in parallel over 21 hours, processing 210 million tokens to examine the data. This intensive analysis led to the identification of over 200,000 reverse transcriptases, which were then narrowed down to 20 compelling candidates.
One Claude agent detected an unusual tandem repeat array in DNA adjacent to a reverse transcriptase, a pattern previously unrecognized as part of a biological system. Claude then quantified the repeats, analyzed their spacing, compared the structure to known reverse-transcriptase systems, and cross-referenced scientific literature. This candidate was subsequently passed to Anthropic's scientists for experimental validation.
Human Expertise Remains Crucial for Validation
A crucial aspect of this discovery is that the process extended beyond data mining and hypothesis generation into experimental validation. Anthropic's human scientists conducted all laboratory experiments and continue to investigate ART's function. Initial findings suggest that the DNA repeat array is transcribed into short RNAs, hinting at a potential functional role similar to repeat-associated components found in CRISPR systems.
Feng Zhang, a pioneer in CRISPR technology and a professor at MIT and the Broad Institute, described the identification of RNA-repeat arrays associated with reverse transcriptases as "intriguing" and deserving of further investigation.
Dario Amodei, Anthropic's CEO, noted that while Claude performed most of the work, human researchers provided initial guidance and executed the laboratory experiments to validate the AI-generated hypotheses. This collaborative workflow underscores AI's capacity to dramatically expand the search for useful biological systems by examining genomic data at a scale unfeasible for individual research teams.
Broader Implications for Biological Research
Anthropic views this discovery as an early indicator of a significant shift in scientific research methodologies. The company is developing a life sciences organization focused on using Claude for a range of tasks, from fundamental biology and genomics to drug discovery. Their Life Sciences Verification Program aims to provide vetted professionals with AI models tailored for biological work.
While the practical gene-editing application of ARTs has not yet been established, the ability of AI to identify such complex biological mechanisms could accelerate the discovery of new drug targets, therapeutic approaches, and improve biological measurement. This advancement shortens the cycle between computational hypotheses and laboratory experiments, potentially transforming the pace of scientific breakthroughs.