AI Discovers DNA System
Anthropic AI system identifies a previously unknown CRISPR-like DNA enzyme system in laboratory experiments.
Anonymous
Key Points
1
ART includes a reverse transcriptase—an enzyme that copies RNA into DNA—and a neighboring partner gene; the reverse transcriptase itself was known previously, while the novelty was recognizing it as part of a larger biological system.
2
Anthropic CEO Dario Amodei told the U.N. Security Council that the preliminary finding “may constitute a new gene editing mechanism that could have applications in gene therapy,” and predicted “profound progress in the near future” on long-standing diseases.
3
Anthropic said the ART system can perform operations resembling “cutting, copying, and pasting” DNA, although the broader research community will need to determine how substantial and novel the discovery is.
4
Dario Amodei acknowledged that the finding builds on earlier work and said a Stanford team had previously discovered a system “that is in some ways similar to the one Claude found.”
5
Feng Zhang, a pioneer of CRISPR genome editing, called the finding “genuinely intriguing” and said it merits further investigation, providing outside scientific support for continued study rather than confirmation of ART as a usable gene-editing tool.
Anthropic's Claude AI system identified a previously unknown enzyme system in bacteriophage DNA, tentatively named array-associated reverse transcriptases (ART), after 950 AI agents analyzed over 200,000 sequences in 21 hours The Hindu Business Line. The system shows structural features similar to CRISPR gene-editing tools, but researchers caution that whether ART can actually target or edit specific DNA sequences remains unknown TechRepublic.
Anthropic CEO Dario Amodei told the UN Security Council the discovery "may constitute a new gene editing mechanism that could have applications in gene therapy" DataConomy. However, independent validation and further laboratory work are needed before ART can be considered a usable gene-editing tool The Decoder.
How Claude Found a Hidden Biological Pattern
Anthropic's 950 AI agents searched through 1.9 billion protein clusters and 200,000 reverse-transcriptase sequences over 21.5 hours TechRepublic. The system narrowed results to 20 candidates, identifying repeating DNA arrays paired with known and uncharacterized proteins. What made the discovery novel: the reverse transcriptase enzyme itself was known, but Claude recognized it as part of a larger three-component biological system including a partner gene and non-coding DNA repeats Times Now News.
Laboratory experiments confirmed the system exists and performs operations resembling "cutting, copying, and pasting" DNA DataConomy. Yet Anthropic acknowledged the finding builds on related prior research, noting a Stanford team had previously discovered a similar system The Hindu Business Line.
Expert Skepticism on Gene-Editing Claims
While Feng Zhang, a CRISPR pioneer at MIT, called the finding "genuinely intriguing" and said it "merits further investigation," he stopped short of endorsing it as a viable gene-editing tool The Decoder. Kevin Blake, a microbiologist at Washington University, stated flatly: there is currently "nothing to indicate the system is a rival to CRISPR as a technology or could be developed into a therapeutic application" TechRepublic.
Lucas Harrington, co-founder of Mammoth Biosciences, dismissed the discovery as "routine automated genome mining" rather than a breakthrough The Decoder. The scientific consensus: identifying structural patterns in DNA databases is different from proving the system can edit genes in living cells.
Amodei Pitches AI Discovery as Global Security Issue
On September 23, Amodei addressed the UN Security Council, framing the ART discovery as proof that AI agents can solve complex biological problems DataConomy. He positioned the find as evidence that AI deserves strict international oversight, urging world leaders to implement safety guardrails against AI-driven bioweapons development DataConomy.
Amodei posted on social media that while biotech usefulness remains unverified, "it is work I would have been proud to do as a PhD student." The announcement raised concerns among AI researchers about promotional framing outpacing peer-reviewed functional confirmation The Decoder.
What Happens Next for the ART System
Anthropic released a technical pre-print describing the discovery, but the broader research community must now verify whether ART exhibits programmable cut-and-paste capabilities or targeted DNA cleavage Times Now News. Wet-lab tests showed RNAs from the ART array comprised up to 8% of total phage RNA within 15 minutes of host infection, a promising sign but not proof of gene-editing function TechRepublic.
The discovery signals a shift toward "hybrid labs," where AI agent swarms generate hypotheses and autonomous agents run experiments, drastically shortening drug research timelines. However, scientists stress that applying ART to actual gene therapy remains highly speculative until further validation occurs The Hindu Business Line.
Sentiment
Mostly PositiveNEGATIVENEUTRALPOSITIVE
Publishers
36
Articles
58
Reach
94
A
AnonymousAuthor
