Claude agents discover a novel CRISPR-like enzyme system
Anthropic reveals its own biology wet lab
Disputed
Importance: major (4 of 5)The takeaway
On September 23, 2026 Anthropic reported that about 950 Claude agents, running for 21 hours on 210 million tokens over a large DNA-sequence database, found array-associated reverse transcriptases (ARTs).
Status
- Claim
Disputed
- Our reporting
- High confidence
- Verification
- Preprint only
- Importance
- Major (4 of 5)
- Last verified
- 9 October 2026
Your AI and this story
- GPT-6 Astra146 days after its cutoff
- Claude Opus 5.585 days after its cutoff
- Gemini 4 Hydrogen176 days after its cutoff
- Grok 4.7115 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 85 days before it.
Key facts
- Announced Sept 23, 2026; technical preprint released
- ~950 Claude agents, 21 hours, 210M tokens
- 200,000+ reverse transcriptases gathered, 3,500 candidate systems, top 20 analyzed
- CRISPR pioneer Feng Zhang (MIT): ‘an exciting example of how AI agents can contribute to biological discovery’
- Anthropic’s wet lab (BSL-1/BSL-2, no human pathogens, all bench work by human scientists) confirmed Sept 18 by head of life sciences Eric Kauderer-Abrams
- Disputed novelty/significance: biologist Lucas Harrington: ‘finding a weird cluster of genes and repeats is often the easy part... the hard part is figuring out what the system actually does’
- Mario Rodríguez Mestre (Univ. of Copenhagen) says his team had already found the pattern and suspects it leaked from his own Claude conversations; Anthropic denies this (says Claude is not trained on user transcripts and its biology team had no access to them). Mestre’s group calls the system “jumbotrons”, first seen in jumbo phages in 2022, still unpublished (NYT 2026-09-27)
- CNN (Katie Hunt, Oct 8): Rodríguez Mestre told CNN ‘It is essentially the same finding’ and ‘I am not claiming that Anthropic deliberately took our work. I cannot demonstrate that’; he thinks it more plausible that ‘there was considerably more prior scientific knowledge and human direction behind the search than the phrase autonomous discovery suggests’. Anthropic did not respond to CNN
Show 1 more
- Other experts in CNN: Aaron Engelhart (Univ. of Minnesota): ‘very interesting preliminary results, but… they don’t demonstrate gene editing or provide a mechanistic picture’; Ilya Finkelstein (UT Austin) noted lead author Peter Yoon is a former Doudna-lab scientist and ‘four of six authors are senior molecular biologists’, and that the case may make scientists wary of putting unpublished results into AI tools; Gustavo Sudre (King’s College London) called the broad-direction search ‘a real shift’ but said ‘the gap between interesting pattern and useful knowledge is where most of the work lives’. The work is not peer-reviewed
What happened
Anthropic formed the life-sciences research group in spring 2026 to test whether general-purpose models can speed up biological discovery. The announcement does not say which Claude model version the agents used.
Why it matters
It is an example of massively parallel agent search yielding a biologically novel finding endorsed by a leading domain expert. It also marks Anthropic’s move into running its own physical experiments.
Sources
15 sources from 13 sites. Numbers match the chips in the text.
15 sources: 3 primary, 9 press, 3 reactions
Primary
- Claude discovers a novel enzyme system with CRISPR-like repeats (Anthropic)anthropic.com, official
- Technical preprint (PDF)www-cdn.anthropic.com, paper
- Anthropic on X: Claude discovers an enzyme systemx.com, official
Press
- The Atlantic (Hana Kiros): AI’s Real Gift to Science, ‘Advanced models can help researchers sort through mountains of biological data’ (Oct 4, 2026; paywalled, only the dek was read)theatlantic.com, press
- TechCrunch: Anthropic says its biology lab has already found something bigtechcrunch.com, press
- TechCrunch: Anthropic is operating a lab that conducts biology experimentstechcrunch.com, press
- SiliconANGLE: Anthropic opens AI-powered biology research labsiliconangle.com, press
- Phys.org: Anthropic touts AI-led biology discoveryphys.org, press
- Irish Times (NYT syndication): Did Anthropic’s AI really make a scientific discovery on its own? (Rodríguez Mestre ‘jumbotron’ priority claim)irishtimes.com, press
- Benzinga: scientist says he had already studied the enzymes for 4 yearsbenzinga.com, press
- The Verge: Anthropic’s biolab: Claude finds a CRISPR-like enzyme systemtheverge.com, press
- CNN: Anthropic’s announcement of a ‘Claude-led’ CRISPR-like discovery is sparking controversy. Here’s whycnn.com, press
Reactions
- MIT Technology Review: When can we say AI made a scientific discovery?technologyreview.com, discussion
- Inside Anthropic’s molecular biology lab (video)youtube.com, video
- Lucas Harrington on X: genome-mining critique threadx.com, discussion
Changes
- Added CNN’s Oct 8 report (Rodríguez Mestre’s own statement; Engelhart, Finkelstein and Sudre)
- Science block: kind: other (not a long-open problem; listed with other AI science results, not in the open-problems count)
- Linked the video “Anthropic says Claude discovered new gene-editing tool as debate over AI in healthcare grows”
- Added The Atlantic’s Oct 4 feature (URL slug and dek point to Anthropic’s biology work; full text blocked, so not checked whether it covers this discovery)
- Added The Verge link
- Added Irish Times/NYT and Benzinga links and ‘jumbotron’ details of the Rodríguez Mestre priority claim; no Mestre preprint or own statement found yet
Show all 9 changes
- Added post link(s) (2) from Anthropic posts cluster
- Added science block and the Harrington / Rodríguez Mestre dispute (MIT Technology Review, 2026-09-28); (science & math tab)
- Filed
