Anthropic Says Claude Helped Find a CRISPR-Like Enzyme System in Its Biology Lab

Anthropic says its Bay Area wet biology lab identified a previously unknown enzyme system hidden in the DNA of bacteriophages, the viruses that infect and replicate within bacteria. The company describes the system as having properties reminiscent of CRISPR, including the ability to perform operations like cutting, copying, and pasting DNA.

The claim matters because CRISPR is widely used by researchers as a gene-editing technology. The next key step is independent validation by the broader research community to assess how significant and novel the discovery really is.

Anthropic CEO Dario Amodei said the discovery was found “mostly, though not entirely, by Claude.” According to the article, Claude’s focused work took 21 hours, using about 950 agents and 210 million tokens to search through data.

That makes the announcement notable not just as a biology claim, but as a demonstration of how AI models may accelerate scientific research workflows. At the same time, Anthropic acknowledged related prior work, including a Stanford-discovered system Amodei described as similar in some ways.

The biggest safety detail is that Anthropic has not let Claude run the wet lab on its own. The article says the physical experiments were performed by human scientists, and Anthropic stated that the lab works only at lower biosafety risk levels, BSL-1 and BSL-2, and does not handle pathogens that can infect humans.

This distinction is important as AI companies debate the risks and benefits of more capable models in sensitive scientific domains. Amodei has warned about AI’s potential use in bioterrorism while also arguing that AI could help cure most diseases in 5 to 10 years.

Anthropic is not alone in applying AI to biological research. The article notes related work from Stanford, UC San Francisco, and Google’s AlphaFold, underscoring that AI-driven biology is becoming a broader research direction rather than a single-company experiment.

Amodei has not ruled out a future in which Claude could autonomously control lab equipment with safeguards, but he said Anthropic is not doing that today. For readers, the takeaway is clear: AI may be moving deeper into biology, but oversight, validation, and lab controls remain central to whether these systems are trusted.