Anthropic Claude Finds CRISPR-Like DNA Pattern But Faces Questions Over Unpublished Research Use

Anthropic Claude Identifies Novel DNA Pattern in Bacteriophages

Anthropic Claude discovered a curious stretch of DNA in bacteriophages that repeats over and over again, similar to CRISPR, the basis of the powerful gene-editing tool CRISPR-Cas91

1

. The newly discovered DNA repeats sit next to a gene for a reverse transcriptase enzyme, a type of enzyme that copies RNA into DNA, suggesting this discovery may have useful properties1

1

. Anthropic named these enzymes array-associated reverse transcriptases (ARTs), presenting the finding as an example of AI initiating biological discovery2

2

. Eric Kauderer-Abrams, Anthropic's head of life sciences, and Nicholas Perry, a researcher at the company, acknowledge it's still unknown exactly what this enzyme and the neighboring repeating pattern of DNA actually do1

1

.

Questions Emerge Over Unpublished Work and Data Privacy

Mario Rodríguez Mestre, a PhD candidate in computational biology at the University of Copenhagen in Denmark, raised concerns about whether Claude independently recognized the significance of these enzymes or if unpublished work shared with the chatbot influenced its result2

2

. Mestre says his team members have been investigating these enzymes, which they call jumbotrons, since 20222

2

. According to The Scientist magazine, he shared his dissertation and manuscript drafts with Claude2

2

. Anthropic denies that the model was trained on user transcripts or that its biology team had access to them2

2

. There is no established evidence that his unpublished material was used in the discovery3

3

. Mestre is named among the inventors of a patent for engineered retrons, genetic systems that contain a reverse transcriptase enzyme and an associated RNA, with its US application published in July 20232

2

.

Understanding the Biological System Behind AI-Generated Findings

What Anthropic claims to have recognized is a larger biological system: the enzyme, a series of repeated DNA sequences, and a possible partner protein3

3

. The researchers found that these repeated sequences produce small RNA molecules3

3

. Nicholas Perry noted that Claude discovered these array of associated reverse transcriptases appear to be isolated to jumbo phages, which have very large genomes for bacteriophage1

1

. Some of that evidence came from reanalyzing data published by another research team in 2022, illustrating how existing data can reveal new findings while highlighting the importance of crediting the scientists who originally produced them2

2

. Reverse transcriptases are enzymes that copy RNA into DNA, and scientists use them in research and diagnostic tests, including some COVID tests2

2

.

Limitations of AI in Scientific Research Revealed

The Anthropic report about Claude's findings explains why checking an AI discovery matters. In 10 additional search runs, the agents missed the repeat array2

2

. Some models could recognize it when given the relevant sequences directly, but recognizing a supplied pattern is different from finding it independently3

3

. This limits claims of a consistently autonomous discovery process by Claude3

3

. Eric Kauderer-Abrams stated that Claude displayed intuition and persistence similar to that of computational biologists, showing exactly the type of logic that really talented computational biologists would go after1

1

. Human scientists conducted the laboratory experiments, and important questions remain unanswered about whether this enzyme is active, whether it copies the associated RNA molecules into DNA, or what the system does in the virus3

3

.

Implications for Future AI Initiating Biological Research

The resemblance to CRISPR-like DNA pattern has attracted attention, but that resemblance does not mean ARTs can edit genes—this remains a possibility to investigate, rather than a proven application3

3

. Before sharing unpublished research with an AI tool, scientists should consult their university's guidance and seek training on responsible use2

2

. Ethical guidelines at institutions like the University of Manchester advise researchers to protect confidential information, verify sources, and keep records of their AI use2

2

. Anthropic's personal and business services have different rules regarding data privacy, with commercial services excluding inputs and outputs from training by default, though exceptions may apply when users give permission or submit feedback2

2

. Watch for how the scientific community establishes standards for verifying AI-generated findings and protecting unpublished work when researchers interact with AI tools.