Anthropic’s AI Claude identifies intriguing DNA stretch in bacteriophages

Anthropic announced on September 23 that roughly 950 autonomous Claude agents ran for 21 hours and screened more than 200,000 reverse transcriptases, flagging about 3,500 candidate systems and narrowing them to 20 for detailed analysis, leading to the discovery of a previously uncharacterized enzyme system in bacteriophage DNA named array-associated reverse transcriptases (ART). The ART system comprises a reverse transcriptase enzyme, an accessory protein of unknown function, and between 3 and 21 evenly spaced non-coding DNA repeats resembling CRISPR arrays, and human scientists at Anthropic's new Bay Area BSL-1 and BSL-2 wet lab validated the computational findings. CRISPR pioneer Feng Zhang called the findings "genuinely intriguing," though no practical applications for ART have been determined and the system's biological function remains unknown; the announcement is the first published result from Anthropic's life sciences research group, established in spring 2026.

Anthropic’s AI Claude identifies intriguing DNA stretch in bacteriophages Nearly 950 autonomous Claude agents scanned hundreds of thousands of genetic sequences to find a previously uncharacterized enzyme system hiding in phage DNA Anthropic https://cryptobriefing.com/markets/anthropic/ just gave its AI model Claude a biology homework assignment, and Claude may have aced it. The company announced on September 23 that a swarm of roughly 950 autonomous Claude agents discovered a previously unrecognized enzyme system lurking in the DNA of bacteriophages, the viruses that infect bacteria. The system has been named array-associated reverse transcriptases, or ART. It consists of three components: a reverse transcriptase enzyme, an accessory protein whose function remains unknown, and a series of evenly spaced non-coding DNA repeats that look strikingly similar to CRISPR arrays. If that last part rings a bell, it should. CRISPR is the gene-editing technology that reshaped modern biology. How Claude found what humans missed The discovery wasn’t the product of a single chatbot session. Anthropic deployed approximately 950 Claude agents that ran autonomously for 21 hours, chewing through around 210 million tokens of genomic data in the process. Those agents screened more than 200,000 reverse transcriptases and flagged roughly 3,500 candidate systems worth a closer look. From that pool, they distilled the findings down to 20 specific systems deemed worthy of detailed analysis and reporting. The reverse transcriptase component of ART wasn’t entirely new to science. Previous studies had identified it within so-called jumbo phages, a class of unusually large bacteriophages. But the full multi-component system, with its accessory protein and CRISPR-like repeat arrays, had never been characterized as a cohesive unit before. Critically, this wasn’t just an AI making bold claims into the void. Human scientists at Anthropic’s newly established wet lab in the Bay Area ran experiments to validate Claude’s computational findings. The lab operates at BSL-1 and BSL-2 biosafety levels and focuses exclusively on fundamental biology research rather than clinical applications. AI, tech, and the markets they move—in one daily briefing. Daily. Free. Join 34,000+ readers across crypto, finance, and policy. What ART actually is, and what it isn’t No practical applications for ART have been determined yet. Nobody knows what the accessory protein does. The biological function of the system as a whole remains a mystery. Feng Zhang, one of the pioneers behind CRISPR gene-editing technology, called the findings “genuinely intriguing” and suggested they deserve further investigation. The ART system features between 3 and 21 DNA repeats arranged in arrays. Anthropic CEO Dario Amodei noted that the discovery builds on prior work connected to research at Stanford University. The announcement represents the first published results from Anthropic’s life sciences research group, which was established in spring 2026 with the explicit goal of using AI for fundamental biology research, hypothesis generation, and experimental validation. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy https://cryptobriefing.com/editorial-policy/ .