Scientists at Stanford University have used artificial intelligence to create new viral genomes that were later tested in a laboratory.
The researchers used an AI model trained on millions of DNA sequences. The study shows that AI can learn patterns from large amounts of genetic information and use those patterns to create new genetic sequences that are different from naturally occurring ones.
According to Forbes, researchers tested the AI-generated sequences in the laboratory, and **16 of them successfully produced functional bacteriophages that could infect E. coli**.
## AI Model Learned From DNA
The researchers used an AI model called **Evo**, which was trained on a very large collection of genomic data.
The approach is similar to how language models learn from large amounts of written text. Instead of learning words and sentences, Evo learned patterns found in DNA and other genetic sequences.
After training, the model was used to generate new genetic sequences. Scientists then synthesized selected sequences and tested whether they could function as viruses.
## 16 AI-Designed Viruses Worked
The researchers found that **16 of the synthesized AI-designed genomes produced functional bacteriophages**.
Bacteriophages, also known as phages, are viruses that infect bacteria rather than humans. In this study, the viruses were tested against **E. coli** in laboratory conditions.
Some of the AI-generated viruses were also able to overcome natural defence mechanisms used by the bacteria.
The results suggest that AI models can learn enough information from genetic data to generate biological sequences that can perform real functions, even when those sequences do not closely match known natural genomes.
## Human Pathogens Were Excluded
The researchers said that data related to human pathogens was excluded from the training process.
This means the study was not intended to create viruses that could infect people. The experiments focused on bacteriophages that target bacteria.
The researchers’ goal was to investigate whether AI could understand biological patterns and help scientists explore new genetic designs.
## Biosecurity Concerns
The findings have also raised concerns among biosecurity experts.
AI systems that can generate functional biological sequences could potentially become useful tools for scientific research and medicine. However, experts warn that the same technology could eventually make it easier to design harmful biological agents if strong safety measures are not maintained.
For this reason, researchers and policymakers are increasingly looking at ways to ensure that powerful AI systems used for biological research include appropriate safeguards.
The Stanford study highlights both the potential and the challenges of using AI in biology. It shows that AI can move beyond simply analyzing existing genetic information and can help generate new biological sequences that work in laboratory experiments.
