Anthropic says its Claude AI model helped identify a novel enzyme system that shares structural patterns with CRISPR, the gene-editing technology, marking what the company describes as the first scientific discovery to come out of its biology research efforts, Times Now reported.
The announcement matters less for novelists directly and more as a marker of how far AI companies are pushing their models beyond writing and coding into original scientific research, a shift that touches every industry watching what these systems can and cannot be trusted to do on their own.
Anthropic said Claude analyzed large DNA databases and spotted an unusual system built around a reverse transcriptase, an enzyme that copies RNA into DNA. The company has named the system array-associated reverse transcriptases, or ART, according to Times Now. It consists of repeating DNA sequences that resemble the repeat arrays characteristic of CRISPR.
The underlying reverse transcriptase enzyme had already turned up in earlier studies, Times Now reported, citing Anthropic. But the company said Claude appears to be the first to recognize the broader system around it, including an array of non-coding DNA sequences and an additional protein whose function is not yet known.
How the search worked
According to Anthropic's account, reported by Times Now, 950 Claude agents searched more than 200,000 reverse transcriptase sequences over 21 hours, processing around 210 million tokens. The agents generated about 3,500 candidate systems and narrowed that list to 20 for detailed investigation.
Blockchain.news reported that one agent flagged a repeating DNA sequence sitting next to a gene for an unusual reverse transcriptase, a pattern reminiscent of CRISPR's repeat arrays. Anthropic's lab experiments then confirmed the presence of an uncharacterized system, though blockchain.news reported that its precise function remains under investigation.
Times Now noted that traditional genome mining requires scientists to manually search large biological databases, spot unusual patterns and formulate hypotheses by hand, a process AI can potentially compress substantially.
Still unconfirmed
The ART system was found in bacteriophages, viruses that infect bacteria, and includes three components: the reverse transcriptase, an accessory protein of unknown function, and the DNA repeat array, blockchain.news reported. Early tests indicated the repeats are expressed as distinct short RNAs, suggesting a possible mechanism similar to CRISPR's RNA-guided DNA targeting.
Feng Zhang, a pioneer of CRISPR genome editing, described the discovery as "genuinely intriguing and merits further investigation," according to blockchain.news.
Blockchain.news reported that no peer-reviewed publication or independent replication has yet verified Claude's findings. Anthropic has released a preprint describing the discovery and has invited outside scientists to collaborate, blockchain.news reported.
Anthropic's approach pairs Claude's hypothesis generation with human scientists who handle all lab-based validation, according to blockchain.news, which also noted that a January 2026 review found AI can accelerate this kind of research but that computational predictions still require rigorous lab testing before their functional significance can be confirmed.
The finding follows Reuters reporting, relayed by Times Now, that Anthropic had quietly built a wet lab in the San Francisco Bay Area as the company expands beyond computer-based work into life sciences and drug-related research.
Whether ART becomes a usable tool for cellular engineering, molecular diagnostics or genetic disease research, as Times Now suggested it might, depends on the peer review and independent replication that neither source reports has happened yet.
Sources
Every quotation and figure in this article is checked against the source reporting linked above before publication. How we report