Scientists Use AI to Design Complete Bacteriophage Genomes, Sparking Biosafety Concerns

Scientists have used AI to design the first complete viral genomes, creating 16 working bacteriophages that kill E. coli. Published in Science, the study used genome language models Evo 1 and Evo 2 to write DNA from scratch. The AI-designed viruses beat antibiotic-resistant bacteria, but experts warn biosecurity governance for generative genome design does not yet exist.

Scientists in the US have used artificial intelligence to design and create viruses that do not exist in nature, in what researchers say is a major step forward for synthetic biology.Researchers at Stanford University and the Arc Institute used a generative AI model called Evo to write new viral genomes. They then synthesised the DNA in a lab and found that 16 of the AI-designed viruses were able to replicate and kill E. coli.The findings were published on Thursday in the journal Science.The viruses were bacteriophages, or phages — viruses that infect bacteria but not humans. The researchers chose them precisely because they are well understood and have been studied extensively.Also Read: AI's future lies in affordable, accessible intelligence, not just smarter models“This is a next step in the complexity that's designable by generative AI,” Brian Hie, a computational biologist at Stanford and the study's lead researcher, told the BBC. “This was new territory for us.”The experiment marks a shift from using AI to design individual proteins or genes to using it to design an entire genome that can actually function inside a cell.Evo works much like a large language model, except that instead of predicting the next word, it learns patterns in DNA. According to The New York Times, the model was trained on genetic sequences from millions of organisms and analysed around nine trillion nucleotides.The researchers then trained it on the 11 genes of Phi X-174, a bacteriophage that infects E. coli, along with about 15,000 of its closest relatives.The model generated hundreds of thousands of possible viral genomes. The researchers selected the most promising candidates for testing and eventually found 16 that produced viable viruses, the New York Times reported.Some of the AI-designed phages were able to multiply faster than the natural Phi X-174 virus.“Our study is a proof of concept showing for the first time that generative design can generate entire functional genomes,” Hie told the Financial Times.That could have major implications for medicine.Phages have long been studied as a potential alternative to antibiotics because they can be engineered or selected to target specific bacteria. With antibiotic resistance becoming a growing problem, AI could make it easier to design phages against particular bacterial infections.“With more evidence coming out on the emergence of scary antibiotic-resistant pathogens, we will need innovative alternative solutions — and phage therapy is certainly one,” Samuel King, a Stanford researcher who worked on the study, told the Financial Times.But this is also where things get uncomfortable.If AI can learn enough about DNA to design a functional virus from scratch, researchers will eventually have to confront what happens when the same technology is applied to viruses that infect humans.Also Read: AMD deepens AI inference bet with Taalas deal as chip race heats upThe Stanford team tried to build safeguards into the experiment. The model was not trained on viruses that infect humans or other complex organisms, while the researchers worked with phages and a non-pathogenic strain of E. coli in a secure laboratory.And unlike many emerging technologies, this one is not confined to a handful of companies. Evo 2 is open source and available to download for free, according to the Financial Times.Hie said the researchers currently have no plans to commercialise the work, arguing that the potential benefits to health and humanity outweigh the risks.The bigger significance of the study may therefore lie beyond the 16 viruses created in a Stanford lab. It shows that AI is beginning to move from designing things on a computer to designing biological systems that can actually work in the real world.And that means the race to figure out what AI can create is now running alongside a much harder race: figuring out what it should be allowed to create.

An anonymous reader quotes a report from The Guardian: Scientists have made the first viruses designed by artificial intelligence in a milestone that raises hopes for new medicines but also concerns over how to ensure the technology remains safe. The viruses are specific kinds known as bacteriophages, which only infect bacteria and are used around the world to treat patients with persistent infections. In lab tests, a cocktail of the AI-designed viruses killed E coli bugs that were resistant to natural bacteriophages. Dr Brian Hie, a chemical engineer at Stanford University in California, used genome language models, the genetic equivalent of the large language models behind AI chatbots, to design functioning genomes for bacteriophages. The viruses were then made in the laboratory and pitted against E coli in a dish. The ability to "rapidly design" genomes and tune them for specific bugs while overcoming resistance could "transform phage therapy" and "expand biotechnological toolkits," the researchers wrote in the journal Science. But beyond the potential benefits, the scientists said the work raised "important biosafety, biocontainment and biosecurity considerations" and urged others who were designing whole genomes to "consult both safety and security professionals throughout the project." In an accompanying article, Prof Tom Inglesby and Dr Moritz Hanke at the Center for Health Security at Johns Hopkins University in Baltimore, reinforced the warning, writing: "Although this is promising for life sciences applications, it also raises urgent biosafety and biosecurity questions. The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not." Tom Ellis, a professor of synthetic genome engineering at Imperial College London, said the work was impressive, but revealed how hard it would be to make more complex genomes. "This is literally the smallest and easiest genome to make," he said. An AI trained on the genetic code of dangerous bugs could be used to design more harmful viruses, Ellis said, but controlling access to genetic data and having restrictions on making genomes that look dangerous would help. "Governments are working hard to do this already," he added. "But honestly," he said, "the threat from full AI design and writing of a genome of a virus or bacteria is very overblown when we consider that just taking existing pathogens and making gain-of-function changes to their genomes is so much easier and much more likely to be a real pathogenic threat." Dr Filippa Lentzos, a reader in science and international security at King's College London, said the most important point to intervene at the moment was when DNA was being manufactured. "It's important to see the bigger governance picture and not focus regulation solely on the AI model," she said. "A layered approach makes more sense: safeguards around model development and access, responsible research review, synthesis screening, and established laboratory biosafety and biosecurity." Read more of this story at Slashdot.