The viruses created during the study are bacteriophages-viruses that infect bacteria rather than humans or animals. Scientists developed 16 new phages capable of attacking E. coli, highlighting AI’s growing role in advancing biotechnology while posing no direct threat to human health.
Experts believe the breakthrough could accelerate the development of treatments for antibiotic-resistant infections, an area of growing concern worldwide. However, the achievement has also sparked fresh debate over the ethical and security implications of using AI to engineer biological systems.
Unlike previous AI applications that focused on discovering new antibiotics, designing an entirely functional virus is significantly more complex. According to Stanford University researcher Brian Hie, the project marks a major step forward because the AI was able to create complete genomes capable of replicating inside living cells.
The research relied on AI models known as Evo1 and Evo2, which operate similarly to large language models. Instead of predicting words and sentences, these systems learn the patterns found in DNA and genetic sequences. The models were trained using genetic information from bacteria, viruses, plants and humans before being refined to design bacteriophages.
After generating hundreds of potential viral genomes, researchers selected 302 candidates for laboratory testing. Sixteen of those designs proved successful, effectively infecting and destroying E. coli bacteria.
For the research team, the first signs of success were unforgettable. Clear patches began appearing on bacteria-covered petri dishes, indicating the newly created viruses were killing the bacteria as intended. The discovery was met with excitement in the lab, with researchers celebrating the achievement after confirming the results.
Scientists say the ability to design customised bacteriophages could eventually transform the treatment of bacterial infections that no longer respond to conventional antibiotics. Phage therapy has long been viewed as a promising alternative, and AI could dramatically speed up the search for effective viral treatments.
At the same time, the study highlights the rapid progress of synthetic biology, where computers are increasingly being used to create biological systems rather than simply analyse them. Researchers believe this technology could lead to the development of new medicines, therapeutic enzymes and improved immunotherapies.
Despite its medical potential, the research has raised concerns among biosecurity experts. In an accompanying commentary published alongside the study in Science, specialists from the Johns Hopkins Center for Health Security warned that advances in AI-driven virus design require immediate attention from regulators and the scientific community.
The experts argued that the question is no longer whether AI can design viruses, but how the technology can be developed responsibly without increasing the risk of misuse. They stressed that creating viruses capable of causing disease should remain off limits.
To minimise risks, the Stanford team deliberately excluded viruses that infect humans and other complex organisms from the AI’s training data. All experiments focused exclusively on bacteriophages and were carried out under secure laboratory conditions.
Researchers involved in the project believe existing safeguards can help ensure the technology is used for beneficial purposes, particularly in medicine and biotechnology.
Although the achievement represents a significant advance, scientists note that designing a living organism remains far more challenging. The genomes of the bacteriophages used in the study contain roughly 5,400 DNA base pairs, whereas even the simplest living cells require around 500,000 base pairs. By comparison, the human genome contains approximately three billion base pairs.
Even so, many researchers see the study as the beginning of a new era. Synthetic biology experts say AI is starting to uncover the principles that govern how genomes are assembled, opening the possibility of designing biological systems on computers before bringing them to life in the laboratory.
While the technology is still in its early stages, its ability to accelerate medical research, improve treatments and deepen our understanding of biology could make it one of the most transformative scientific developments of the coming decade-provided it is guided by strong ethical and safety standards.
No newer articles found.