Science & Technology: Artificial intelligence is entering a new and highly sensitive area of biological research, with scientists demonstrating that AI can help design previously unknown viruses that target bacteria. The development could open new possibilities for fighting antibiotic-resistant infections, but experts are also warning about the potential risks associated with increasingly powerful biological design tools.
According to MIT biologist and associate professor Kevin Esvelt, scientists have now generated 16 previously undetected viral genomes using artificial intelligence. These viruses were designed to infect bacteria rather than humans, making the research particularly relevant to the search for alternatives to antibiotics.
The concept is based on bacteriophages, viruses that naturally infect bacteria. Scientists have explored bacteriophages for decades as a possible way to target harmful bacteria, particularly as antibiotic resistance makes some infections increasingly difficult to treat.
Traditional phage research often requires scientists to search for suitable viruses in nature. However, naturally occurring viruses may not work against bacteria that have already developed resistance to them.
AI-assisted design could potentially expand the range of viruses researchers can study. Instead of relying only on viruses found in nature, scientists can use computational systems to explore new biological possibilities and identify candidates that could potentially target specific bacteria.
This creates an important medical opportunity. Antibiotic-resistant infections are a growing global health challenge, and researchers are looking for new methods of killing harmful bacteria without relying exclusively on conventional antibiotics.
However, the same technology that could help scientists develop new treatments also raises concerns about biosecurity.
Esvelt has warned that the ability to use AI for biological design could potentially be misused. As AI systems become more capable of designing biological sequences, researchers and policymakers face the difficult task of encouraging beneficial research while preventing dangerous applications.
The issue is particularly sensitive because biological information can potentially be designed and tested much faster than in previous decades. This means that safety systems, oversight and responsible research practices need to keep pace with technological progress.
The recent work does not mean that AI has created viruses capable of infecting humans. The reported viruses were designed to target bacteria, and the potential medical application is primarily connected to combating bacterial infections.
Experts therefore see the development as both an opportunity and a warning. If carefully controlled, AI-assisted biological design could contribute to new treatments for infections that no longer respond well to existing antibiotics. At the same time, researchers say safeguards are needed to reduce the possibility of misuse.
The development highlights a broader challenge facing modern science: technological advances can create powerful tools for improving human health while simultaneously introducing new risks.
For medicine, the potential benefit is significant. AI could eventually help researchers identify highly targeted biological therapies and accelerate the search for alternatives to traditional antibiotics. But realizing that potential will require careful laboratory testing, scientific oversight and strong biosecurity measures.
The debate is therefore no longer simply about whether AI can design new biological systems. It is increasingly about how society can use those capabilities safely and responsibly.
