Scientists Use AI to Create New Viable Virus, Sparking Biosafety Debate
Scientists have for the first time successfully used an artificial intelligence (AI) model to create a new viable virus that has never been seen in nature. The breakthrough, while medically promising, has sparked fresh concerns about the potential misuse of AI technology in creating dangerous pathogens.
The study, published in the journal Science, details the creation of a virus that specifically targets bacteria and is harmless to humans. This type of virus, known as a bacteriophage, represents an important medical intervention for fighting drug-resistant pathogens.
“This is an important milestone,” said Patrick Cai, a synthetic biologist at the University of Manchester who was not involved in the study, as quoted by The New York Times.
Researchers from Stanford University and the Arc Institute trained an AI model called Evo to recognise patterns of genetic structure in nature in order to create new DNA molecules. These molecules were then inserted into bacteria, which over time transformed into entirely new viruses that killed their hosts.
Unlike general-purpose large language models (LLMs), Evo was trained on billions of lines of genetic sequences, including millions of bacteriophage genomes. However, its workings are similar to a chatbot; Evo predicts which DNA building blocks (nucleotides) should follow one another, much like AI predicts words to complete a sentence.
In their experiments, the researchers focused on an E. coli-killing virus called Phi X-174. Using Evo, they generated 700,000 candidate new viruses. After narrowing these down to 300 candidates for reconstruction in the laboratory, 16 proved viable and capable of infecting and killing their bacterial hosts.
Despite the clear medical benefits, the ability of AI to assemble viral genomes raises urgent questions about biosafety. Tom Inglesby and Moritz Hanke from the Center for Health Security at Johns Hopkins University warned that governance to safely direct this technology is not yet available.
However, some experts believe the threat may be overstated for now. Tom Ellis, professor of synthetic genome engineering at Imperial College London, said that creating more complex genomes would be far more difficult.
“The threat from full AI design and writing of viral or bacterial genomes is greatly exaggerated when we consider that taking an existing pathogen and making gain-of-function changes to its genome is far easier and more likely to be a real pathogenic threat,” Ellis explained.
At present, controlling access to genetic data is considered one of the key steps to curb the potential threat from AI-based virus engineering in the future.
Source: jawawa.id (https://jawawa.id/newsitem/scientists-use-ai-to-create-new-viable-virus-sparking-biosafety-debate-1786121490)
Originally published by Media Indonesia (https://mediaindonesia.com/internasional/919564/ilmuwan-gunakan-ai-untuk-ciptakan-virus-baru-yang-tidak-ada-di-alam), translated from Indonesian