Want to bookmark your favourite articles and stories to read or reference later? Start your Independent Membership today.
Already a member?
Log in
Anthropic says that its Claude artificial intelligence system has found something potentially transformative in the human body.
The company has found a “previously unknown enzyme system” inside DNA that that it likened to CRISPR, a biotechnology tool that has revolutionised biological research and the discovery of which won the Nobel prize in 2020. But it also warned that it will “take much more work to learn what this system does, and whether it can be put to similar use”.
The announcement of the discovery also marked the company’s first public acknowledgement of a biological lab that it has opened in California. That allows Anthropic to allow Claude to generate biological hypotheses and then test them in real life.
Anthropic says that it will allow it to test the use of artificial intelligence in biological research. But it has also further added to fears about the potential dangers of AI, many of which relate to its potential use to create biological weapons.
The company stressed that its research “involves only the lower-levels of the biosafety risk level” and that it does not “handle pathogens that can infect humans”. “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,” said Dario Amodei, Anthropic’s chief executive.
For now, much of the lab’s work involves setting Claude to search through DNA sequences that are associated with proteins whose function scientists don’t understand. Claude reads the scientific work on the topic, writes a report on it for the humans that run the experiments, and then evaluates the evidence, dismissing most candidates.
If something survives this review process, researchers then test it in Anthropic’s lab, examining the protein so that it can be characterised, which also uses Claude. Scientists can then try and understand if that protein might be useful for further scientific research.
Now, that process has led to its first public announcement. Mr Amodei said that it had turned up “a molecular machine that we suspect could represent a new gene editing mechanism”.
“Its precise function, biotechnological utility (if any), or level of significance is not yet clear, but at minimum it is work I would have been proud to do as a PhD student,” he wrote. “The work was done mostly, though not entirely, by Claude: our life sciences team suggested a broad area of research, Claude read through the literature and a bunch of genome data and discovered something interesting, then Claude proposed experiments to verify the discovery and our team carried them out.”
Many researchers pointed out that the company had only announced that it had made a potentially interesting discovery, and that further work would be required to understand if it was actually useful. Feng Zhang, one of the researchers who pioneered CRISPR, said for instance in a quotation shared by Anthropic only that it was an “intriguing” finding and that further work was required.
“This is an exciting example of how AI agents can contribute to biological discovery,” he said. “The identification of RNA-repeat arrays associated with reverse transcriptases is genuinely intriguing and merits further investigation.”