Scientists at Stanford University have successfully employed artificial intelligence to design novel viral genomes that function and replicate in laboratory settings. The research represents the first instance of complete genomes being designed through generative AI technology. The 16 resulting viruses were engineered to infect bacteria and pose no threat to humans, addressing concerns about potential harm from the technology.
The AI models, designated Evo1 and Evo2, operate on principles similar to large language models such as ChatGPT, but predict genetic sequences rather than text. These models were trained on genetic codes derived from viruses, bacteria, plants, and humans before being refined to produce bacteriophages—viruses that target specific bacterial species. Researchers selected 302 promising AI designs for laboratory synthesis, and 16 proved effective at infecting and killing E. coli bacteria.
The breakthrough carries significant implications for treating antibiotic-resistant infections and developing new therapeutic approaches. Researchers highlight the potential to “massively improve human health” through development of novel drugs and therapies using synthetic biology principles. However, the achievement has simultaneously raised “urgent” biosafety and biosecurity concerns among health experts, who note that similar technology could theoretically be misused to create harmful pathogens.
The Stanford team implemented multiple safety measures during their research, including excluding viruses capable of infecting complex organisms from their training data, conducting experiments exclusively on bacteriophages rather than human pathogens, and performing all work in secure laboratory facilities. Researchers emphasized that existing safeguards help ensure responsible application of the technology. While viruses are not classified as living organisms, experts acknowledge that designing living cells through AI would require additional significant advances, though some researchers indicate interest in pursuing simpler organisms in the future.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI