Claude Has Identified New Biological System with CRISPR-like Features

Anthropic says its Claude artificial intelligence model had identified a previously unknown biological system in the DNA of bacteriophages, viruses that infect bacteria, in an early result from the company’s new life sciences research program.

The system, which Anthropic has named array-associated reverse transcriptases, or ART, has some structural similarities to CRISPR, the natural bacterial defense system that has become a widely used gene-editing technology.

Anthropic said it does not yet know what ART does, meaning it is too early to say whether it could have applications similar to CRISPR. The company said its first experiments showed that ART contains repeating DNA sequences and produces short RNA molecules, but further work is needed to establish its biological function.

The finding is also an early example of Anthropic’s effort to use AI agents to conduct biological research with limited human direction.

Anthropic said about 950 Claude agents spent 21 hours examining a large database, using about 210 million tokens. The agents analyzed more than 200,000 reverse transcriptases, a type of enzyme that copies RNA into DNA, identified 3,500 candidate systems and narrowed them to 20 candidates for detailed analysis.

One agent noticed a repeating DNA pattern near a reverse transcriptase and investigated it further. It compared the pattern with known biological systems, searched scientific literature and eventually flagged it for human researchers to review.

The reverse transcriptase itself was not new. Anthropic said it had previously been identified in a jumbo bacteriophage. What the AI system identified was the combination of the enzyme with a neighboring gene and a long sequence of repeated DNA.

The researchers named the resulting system ART. It consists of three main components: a reverse transcriptase, a neighboring partner gene and a long array of repeated DNA sequences.

The repeated sequences resemble those found in CRISPR systems. In CRISPR, such sequences form part of a system that can be programmed to recognize specific genetic material.

Anthropic said early laboratory experiments indicate that ART’s repeated DNA sequences are also transcribed into separate short RNA molecules. However, it has not established whether ART can target or modify DNA in the way CRISPR-based systems can.

Feng Zhang, a pioneer of CRISPR gene editing and a professor at the Massachusetts Institute of Technology and the Broad Institute, said in comments included by Anthropic that the finding warranted further investigation.

“This is an exciting example of how AI agents can contribute to biological discovery. The identification of RNA-repeat arrays associated with reverse transcriptases is genuinely intriguing and merits further investigation. I hope this work encourages more scientists to explore how AI can support their research.”

The research illustrates a potential shift in how AI could be used in science. Rather than simply analyzing data or answering questions posed by researchers, AI agents could search large datasets, decide which leads to investigate and generate candidates for laboratory testing.

Anthropic said its researchers still carried out the experiments needed to test Claude’s findings. The company is also exploring ways to connect AI systems with laboratory equipment.

Anthropic Chief Executive Dario Amodei said in a post on X that eventually Claude could potentially perform experiments itself by controlling laboratory equipment, subject to appropriate safety measures.

The company said the biological function of ART remains under investigation and that not every candidate identified by Claude will lead to a scientific discovery.

The work is part of a new life sciences laboratory established by Anthropic to use Claude to search biological datasets, generate hypotheses and test promising candidates experimentally.

The results have not established that ART is a new gene-editing technology. Instead, they show how AI agents can help researchers identify previously uncharacterized biological systems from very large datasets, Anthropic said.

Stay ahead of the Stories shaping our world. Subscribe to Impact Newswire and join our 
WhatsApp Channel for updates on global tech, business, and innovation—all in one place.

Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.

Got a story to share? Contact Us to reach a global audience with Impact Newswire.