Researchers create AI-designed viruses that could take on “cabbage black rot,” among other bads

Stanford University researchers have successfully created functional viruses using AI-designed genomes, marking what they call “the first generative design of complete genomes.” The breakthrough demonstrates AI’s potential to design biological systems from scratch, opening new possibilities for treatments while raising important biosecurity concerns.

How it works: Scientists trained an AI model called Evo on the genomes of about two million bacteriophage viruses to learn genetic patterns.

• The researchers focused on phiX174, a simple virus with only 11 genes and about 5,000 DNA letters that infects bacteria.
• They used the same underlying technology as ChatGPT, but instead of training on text, the AI learned from viral genetic sequences.
• The team then chemically printed 302 AI-generated genome designs as DNA strands and tested them with E. coli bacteria.

The results: Out of 302 AI-designed viral genomes tested, 16 successfully replicated and killed bacteria.

• “That was pretty striking, just actually seeing, like, this AI-generated sphere,” says Brian Hie, who leads the lab at the Arc Institute, a non-profit research organization in Palo Alto where the work was carried out.
• The successful viruses appeared as plaques of dead bacteria in petri dishes, with microscope images showing tiny viral particles that look like fuzzy dots.
• The AI created unexpected variations, including “viruses with new genes, with truncated genes, and even different gene orders and arrangements,” according to Jef Boeke, a biologist at NYU Langone Health.

Why this matters: The achievement represents a significant step toward AI-designed life forms and could accelerate development of new treatments.

• Computer-designed viruses could enhance “phage therapy” for treating serious bacterial infections and agricultural diseases like cabbage black rot.
• Most gene therapy relies on viruses to deliver genes into patients’ bodies, and AI might develop more effective delivery systems.
• The speed advantage over traditional methods is crucial—J. Craig Venter, who created some of the first organisms with lab-made DNA, notes this looks like “just a faster version of trial-and-error experiments” compared to his team’s manual approach in 2008.

Current limitations: While promising, the technology faces significant scaling challenges for larger organisms.

• Viruses aren’t technically alive and have relatively simple genomes compared to bacteria or complex life forms.
• E. coli has about a thousand times more DNA code than the phiX174 virus used in this study.
• “The complexity would rocket from staggering to … way way more than the number of subatomic particles in the universe,” says Boeke about designing larger genomes.

Biosecurity concerns: The technology raises important safety questions about potential misuse.

• The Stanford team purposely avoided training their AI on viruses that infect humans.
• “One area where I urge extreme caution is any viral enhancement research, especially when it’s random so you don’t know what you are getting,” warns Venter.
• “If someone did this with smallpox or anthrax I would have grave concerns.”

What’s next: Industry leaders see this as a stepping stone toward fully automated biological design.

• Jason Kelly, CEO of Ginkgo Bioworks, a cell-engineering company, envisions “automated” laboratories where genomes get proposed, tested, and results fed back to AI for improvement.
• “This would be a nation-scale scientific milestone as cells are the building blocks of all life,” says Kelly. “The US should make sure we get to it first.”
• The work builds on AI’s growing impact in biology, following the 2024 Nobel Prize for AI protein structure prediction and recent major investments like Boston company Lila’s $235 million raise for AI-driven labs.