Short answer
Anthropic says its Bay Area biology lab has found a new enzyme system with properties similar to the gene-editing system CRISPR. According to the company, the discovery was largely made by its AI model Claude, while human scientists carry out the physical experiments. It shows large AI companies starting to bring physical laboratories into their research process.
Highlights
- Anthropic says Claude largely made the discovery; the process took 21 hours and used around 950 AI agents.
- A total of 210 million tokens were used in the process.
- Amodei says he believes AI could cure most diseases within 5-10 years.
2 min read
Anthropic has announced, as reported by TechCrunch, that it has discovered a previously unknown enzyme system in its wet biology lab in the Bay Area. The company says the system, found in bacteriophage DNA, has properties similar to CRISPR.
CRISPR is a natural immune mechanism that bacteria developed against viruses and that researchers turned into a gene-editing technology. According to Anthropic's researchers, the newly found system can also perform operations such as cutting, copying and pasting DNA.
Did Claude really make the discovery?
According to Anthropic, the discovery was made largely, but not entirely, by Claude. The company says the process took 21 hours, with around 950 AI agents scanning the data and a total of 210 million tokens used. Anthropic CEO Dario Amodei acknowledges that a team at Stanford had previously found a similar system and that this discovery also builds on the work of others.
Who works in the lab: humans or AI?
Anthropic stresses that the lab works only at low-risk biosafety levels and does not handle pathogens that can infect humans. According to the company, all physical experiments are carried out by human scientists; for now, Claude does not directly control lab equipment. Amodei also says he believes AI could cure most diseases within the next 5-10 years, but that the risk of AI being misused for bioterrorism is also one of the things he fears most.
What does this signal?
This development shows that large AI companies are starting to bring physical laboratories into their research process. Similar work at universities such as Stanford and UC San Francisco indicates that this is not unique to a single company.
Frequently asked
- Does this mean AI is working fully autonomously in the lab?
- No. Anthropic states clearly that Claude runs the data analysis, but the physical experiments are done by human scientists.
Sources
Follow UNIT Journal
What's new in search, AI and technology, in your feed every day.
