Stanford researchers used artificial intelligence to design complete viral genomes that produced 16 laboratory-reproducible bacteriophages targeting E. coli, a scientific first that could aid efforts against antibiotic resistance while intensifying debate over the potential misuse of AI-generated biology.
By yourNEWS Media Newsroom
Researchers at Stanford University have demonstrated that generative artificial intelligence can design complete viral genomes capable of producing functioning, replicating viruses, a development that could eventually contribute to new treatments for antibiotic-resistant infections but is also drawing warnings about the biosecurity implications of increasingly powerful biological AI systems.
The experiment produced 16 new bacteriophages, viruses that infect bacteria rather than people or animals, after researchers used AI models to generate hundreds of potential designs targeting E. coli. According to BBC reporting on the research, the scientists relied on models known as Evo1 and Evo2, which operate differently from text-generating systems such as ChatGPT. Instead of predicting words, the models were trained to predict genetic sequences found in viruses, bacteria, plants and humans. The researchers then used those capabilities to propose entirely new bacteriophage genomes, 16 of which could be successfully produced in the laboratory. The findings were published in the journal Science.
The work represents the first successful use of generative AI to design an entire genome, moving the technology beyond the creation or modification of individual biological components and into the construction of genetic systems capable of replication and activity inside cells. Brian Hie, an assistant professor at Stanford University, said the experiment marked a significant expansion in what AI can biologically design. “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,” Hie told the BBC.
One potential application is bacteriophage therapy. Because bacteriophages attack bacteria, scientists have long studied whether they could be used against infections that have become resistant to conventional antibiotics. The ability to generate new phages with AI could eventually give researchers another way to develop treatments against difficult bacterial infections, particularly as antibiotic resistance reduces the effectiveness of existing drugs.
The same technical advance, however, is prompting questions about how far generative biological design should be allowed to progress. Synthetic biology involves engineering biological systems and, in some cases, creating forms of life that do not naturally exist. As AI systems become capable of designing increasingly complex genomes, biosecurity specialists are concerned that similar methods could eventually be directed toward pathogens capable of harming humans.
In commentary published alongside the study in Science, Dr. Thomas Inglesby and Dr. Moritz Hanke of the Johns Hopkins Center for Health Security said the experiment creates “urgent biosafety and biosecurity questions.” They argued that the central question has already moved beyond “whether generative viral genome design will exist” and instead concerns whether such technology can be developed and deployed without “enabling serious harm.” The researchers specifically warned that attempts to engineer novel viruses capable of causing disease “should not be pursued.”
Hie said the Stanford team attempted to build safeguards into the project before generating the experimental viruses. Viruses capable of infecting complex organisms were excluded from the training database, he said, and researchers limited their work to bacteriophages that infect bacteria rather than people or animals. The laboratory work was also conducted in a secure environment. Hie said those precautions and other existing safeguards can help ensure “that the technology is used for good.”
The experiment nevertheless illustrates how quickly artificial intelligence is expanding from software-based applications into biological engineering. A system capable of predicting genetic code can potentially be used not only to analyze existing organisms but also to propose biological designs that have never existed before. In this case, the resulting organisms were bacteriophages directed against E. coli, and the project was undertaken with potential medical uses in mind. Biosecurity experts are warning that the underlying capability could become more consequential if similar systems are eventually able to design dangerous viruses or other pathogens.
That dual-use potential places the Stanford findings at the intersection of medical innovation and security policy. The same ability to rapidly design bacteriophages could help researchers develop new weapons against antibiotic-resistant bacteria, while continued improvements in genome-generating AI could make biological design increasingly accessible and sophisticated. Inglesby and Hanke’s warning reflects concern that safeguards will need to advance alongside the technology rather than after more dangerous capabilities have already become available.
Additional details about the experiment and its implications are available through the BBC’s report, which describes both the potential medical applications and the growing debate over whether current biosecurity measures are sufficient for a new generation of AI systems capable of designing entire viral genomes.
Original article
Comments