Anthropic confirmed last week that it operates a wet biology lab in the Bay Area, where it uses AI models to conduct physics experiments. This week, the AI giant announced that the lab has already achieved what it believes to be a major discovery: a new enzyme “system” that has “certain properties reminiscent of CRISPR,” as Anthropic describes it.
CRISPR is the innate immune system that bacteria use to fight viruses and has become a widely used gene editing technique among researchers.
Essentially, what Anthropic says it has discovered is a previously unknown enzyme system hidden in the DNA of bacteriophages. Bacteriophages are viruses that infect and replicate within bacteria. Anthropic researchers say the system works similarly to CRISPR in that it “can perform operations such as cutting, copying, and pasting DNA.”
It will be up to the broader research community to verify how significant or new this discovery really is. Anthropic CEO Dario Amodei wrote in X that the discovery was based on the work of others, and acknowledged that a team at Stanford University had previously discovered a system “similar in some ways to what Claude discovered.”
But Amodei and Anthropic aren’t just highlighting this finding. They tout it as being discovered “mostly, but not entirely, by Claude,” the CEO writes.
On the one hand, it’s amazing. The institute was founded just this spring (though Anthropic declined to provide details on how many months ago it has been open). While the discovery was quick enough in just a few months, the researchers say Claude actually worked together for only 21 hours. Claude used around 950 agents to search the data and burnt 210 million tokens.
Still, the revelation that Anthropic was conducting biology experiments came shortly after AI CEOs, including Amodei, publicly acknowledged that models had become so capable and potentially so dangerous that the industry needed to slow down and develop safety testing procedures. This comes after several of his employees publicly said earlier this month that AI risks killing us all.
Amodei himself has said that one of his biggest fears is that AI could be used for bioterrorism. But he also wrote that he believes AI will “cure most diseases within five to 10 years.” Therefore, Anthropic clearly determined that the reward was worth the risk.
With that in mind, perhaps the biggest revelation from this news is not the discovery itself, but that the biology lab is not letting Claude escape. Physics experiments were performed by humans.
“Our lab in the Bay Area looks like a typical molecular biology lab. We conduct research involving only low biosafety risk levels (BSL-1 and BSL-2) and do not work with pathogens that can infect humans. All lab work is performed by human scientists.”
It’s also true that AI-powered biological research is not unique to Anthropic. Researchers at Stanford University have published a paper on the work they are doing using LLM and CRISPR. Researchers at the University of California, San Francisco used AI to design an enzyme from scratch. In 2020, Google announced AlphaFold, the first AI-powered biological research tool. So AI has clearly reached the field of biological research beyond the AI labs themselves pursuing it.
However, Amodei does not rule out the possibility of fully automated biology laboratories in the future. “Eventually, it may even be possible for Claude to conduct experiments safely himself by autonomously controlling the experimental equipment with appropriate safety measures in place, but we are not doing that at this time,” he said.
.
If you make a purchase through links in our articles, we may earn a small commission. This does not affect editorial independence.
