Anthropic reports biology discovery made with AI assistance

Anthropic has announced that its artificial intelligence model, Claude, helped discover a new enzyme system with properties similar to gene-editing technology.

2 min read

Line drawing of a human profile with a neural network symbol near the head on an orange background.
An artistic illustration representing artificial intelligence and biological research connections.

Anthropic announced that its wet biology laboratory in the Bay Area has helped discover a previously unknown enzyme system. The system was identified within the DNA of bacteriophages, which are viruses that infect bacteria, and exhibits properties reminiscent of CRISPR, a natural immune system used by bacteria that has become a widely used gene-editing technology.

Key facts

Research group formation
Spring of 2026
Search time
21 hours
Number of agents
Roughly 950 agents
Tokens used
210 million tokens

The discovery process

According to the company, the discovery was made largely through the efforts of its Claude artificial intelligence model. The model searched through biological data using approximately 950 agents, consuming 210 million tokens during 21 hours of concerted effort.

Chief Executive Dario Amodei noted that the work builds on prior research, including similar systems previously identified by researchers at Stanford University. The broader scientific community will need to review and validate the novelty of the finding.

Safety and human oversight

While the artificial intelligence model analyzed data and directed research directions, all physical experiments in the laboratory were conducted by human scientists. The facility operates at the lower levels of biosafety risk, designated as BSL-1 and BSL-2, and does not handle pathogens capable of infecting humans.

Anthropic noted that while it does not currently allow artificial intelligence models to autonomously control laboratory equipment, such capabilities might be explored in the future under appropriate safeguards.

Context in biological research

The intersection of artificial intelligence and biology is an active field. Other institutions, including Stanford University and the University of California, San Francisco, have recently published work involving artificial intelligence and enzyme design, building upon foundational tools like DeepMind's AlphaFold.

Source: Original announcement

Worth a read?

Report

Comments