Anthropic Says Bay Area Biology Lab Has Found CRISPR-Like Enzyme System

Anthropic confirmed last week that it operates a wet lab where it uses AI models to run physical experiments, TechCrunch reported. This week, it announced the lab had already produced what it considers a major finding: an enzyme system with properties the company compares to CRISPR, the natural bacterial immune system that researchers have adapted into a widely used gene-editing technology. The system was found in bacteriophage DNA. Bacteriophages are viruses that infect and replicate within bacteria. Anthropic researchers said the system can perform operations such as cutting, copying and pasting DNA.

The significance and novelty of the finding will need validation from the broader research community. Amodei acknowledged on X that the work built on previous research and that a Stanford team had earlier discovered a system 'that is in some ways similar to the one Claude found,' TechCrunch reported. Even so, Amodei and Anthropic emphasized that the system was found mostly, though not entirely, by Claude.

The lab was established this spring, and Anthropic declined to say more specifically how many months it has been open, according to TechCrunch. The company said Claude spent 21 hours of concentrated effort on the search, using about 950 agents and consuming 210 million tokens. A discovery within a few months would already be fast, but Anthropic attributes much of the work to the model rather than to human researchers.

The announcement comes as AI chief executives, including Amodei, have publicly said that models have become so capable and potentially dangerous that the industry must slow down and develop safety-testing procedures. Earlier this month, a couple of Anthropic employees publicly said there is a risk AI could kill all humans, according to TechCrunch. Amodei has said one of his biggest fears is that AI could be used for bioterrorism. He also wrote that he believes AI will 'cure most diseases in 5-10 years.' TechCrunch reported that Anthropic has decided the rewards are worth the risks.

For now, the biology lab has not let Claude operate freely. Anthropic said the physical experiments were done by humans. 'Our lab, located in the Bay Area, looks like a typical molecular biology lab. We do research that involves only the lower-levels of the biosafety risk level (BSL-1 and BSL-2) and we do not handle pathogens that can infect humans. All of the lab work is performed by human scientists,' the company said, according to TechCrunch.

AI-powered biological research is not unique to Anthropic. Stanford researchers have published a paper on work with LLMs and CRISPR, and researchers at UC San Francisco used AI to design enzymes from scratch, TechCrunch reported. Google launched AlphaFold, its first AI-powered biology research tool, in 2020. AI has therefore arrived in biological research well beyond the AI labs pursuing it themselves.

Amodei has not ruled out a fully automated biology lab in the future. 'Eventually it may even be possible for Claude itself to safely perform the experiments by autonomously controlling lab equipment, with appropriate safeguards in place, but we aren't doing that today,' he said, according to TechCrunch.

Editor's Summary

Anthropic says its Bay Area wet lab found a CRISPR-like enzyme system in bacteriophage DNA, mostly through Claude, while humans performed all physical experiments. The company says the finding still needs outside validation and builds on earlier work by others. The announcement comes amid debate over AI safety and biological risks, though Anthropic has not yet allowed Claude to run lab experiments autonomously.