Scientists have created functional viruses designed with artificial intelligence, marking a major development in AI-powered biotechnology.
The breakthrough also raises important questions about biosafety, biocontainment and biosecurity as AI capabilities expand.
The viruses are bacteriophages, which infect bacteria rather than humans. Some parts of the world already use bacteriophages to treat persistent bacterial infections.
In laboratory tests, a combination of the AI-designed viruses successfully killed E. coli bacteria resistant to natural bacteriophages.
AI Designs Complete Viral Genomes
Dr Brian Hie, a chemical engineer at Stanford University, and his team used genome language models to design functioning bacteriophage genomes.
These systems process genetic sequences in ways that resemble how large language models process and generate text.
The researchers then produced the designed viruses in a laboratory and tested them against E. coli bacteria.
According to the researchers, the technology could allow scientists to design genomes rapidly for specific bacteria.
It could also help researchers overcome resistance to existing bacteriophages, potentially expanding the tools available for phage therapy.
Why AI-Designed Viruses Matter
The research demonstrates how AI can move beyond analysing biological information and help generate functional genetic designs.
That capability could create new opportunities for biotechnology and medical research, particularly in areas involving bacterial infections.
However, the same development has also intensified concerns about how such technologies should be developed and controlled.
The researchers therefore urged scientists working on whole-genome design to involve safety and security experts throughout their projects.
Scientists Warn of Biosafety Risks
Prof Tom Inglesby and Dr Moritz Hanke of the Johns Hopkins Center for Health Security warned that generative AI can now compose functional viral genomes.
At the same time, they noted that governance mechanisms have not yet caught up with this capability.
Tom Ellis, professor of synthetic genome engineering at Imperial College London, described the research as impressive.
However, he pointed out that bacteriophages have relatively small and simple genomes.
Designing more complex organisms would therefore present considerably greater challenges.
Ellis also warned that AI trained on genetic information from dangerous organisms could potentially help design more harmful viruses.
Nevertheless, he said restricting access to sensitive genetic data could reduce those risks.
Screening suspicious genome synthesis requests could provide another layer of protection against misuse.
Ellis also argued that modifying existing pathogens remains an easier and more immediate threat than creating an entirely dangerous genome using AI.
Calls Grow for Layered AI-Biosecurity Controls
Dr Filippa Lentzos of King’s College London said regulation should not focus solely on AI models. Instead, she called for a layered approach covering several stages of the technology’s development and use.
That approach should include model development and access, research oversight and DNA synthesis screening. It should also include established laboratory biosafety and biosecurity procedures.
The concerns reflect the broader challenge created by rapidly advancing AI capabilities in biotechnology. As researchers gain greater ability to design functional biological systems, safety measures will need to develop alongside the technology.
For now, the creation of functional AI-designed bacteriophages represents both a scientific milestone and a warning.
The technology could help researchers address difficult bacterial infections. At the same time, it highlights the need for careful oversight as AI enters increasingly powerful areas of biological research.
