For the first time, scientists are using artificial intelligence, or AI, to design new kinds of viruses, pushing breakthroughs in medicine while also raising critical biosecurity concerns. As computer programs push the boundaries of biology, researchers and policymakers are trying to balance this innovation with emerging risks.

Instead of using traditional genetic editing or investigating antiviral drugs, researchers at Stanford University and the Arc Institute trained a genomic language model, called Evo, on trillions of genetic letters. DNA is a long molecule made up of chemical units called nucleotides which are represented by the letters A, C, G, and T. This series of letters stores the instructions for creating proteins and other molecules. Additionally, DNA has its own mysterious rules to produce a working protein or biological outcome, and if those rules are broken, the sequence would not produce a working biological structure. Biologists have uncovered some of their rules; however, many still remain undiscovered. Now, researchers wonder if Evo could pick up these rules on its own. Over time, by mastering genetic patterns across the tree of life, Evo successfully generated blueprints for multi-job proteins, prompting researchers to test whether the AI could design an entire genome from scratch. Scientists chemically synthesized around 300 candidates in the lab, ultimately producing 16 fully functional live viruses that have never existed in nature. “This is a next step in the complexity that’s designable by generative AI. This is the first time generative AI has been used to design a complete genome, it’s something that can replicate and have other functions inside cells… this was new territory for us,”said Brian Hie, an assistant professor at Stanford University.

The researchers have reached a very important milestone in medical science. The researchers created the DNA molecules, which were inserted into bacteria. Later, the newly modified bacteria produced viruses never seen in nature. The viruses were able to infect other bacteria, proving that they were viable. Most importantly, those viruses created by AI are not a threat to humans because they are similar to a natural virus called Phi X-174, which can only infect bacteria. Those AI-designed viruses offer a powerful new way to destroy antibiotic-resistant bacteria like E. coli without harming human cells. This allows the scientists to quickly build treatments.

While the researchers took strict safety measures, excluding human pathogens from the AI’s data and working in high security facilities, this study poses a serious risk. Since model architectures like Evo are freely accessible, experts worry that dangerous people could eventually apply similar methods to design dangerous human pathogens or other weapons, such as deadly poisons or incurable pandemics. Society now faces a serious danger. The technology to compose viral genomes exists, but global regulations to monitor it still have to catch up.

The future of AI generated biology is still being discovered, and its success is based on how fast researchers can establish safeguards to control these medical breakthroughs without creating new threats.