Artificial Intelligence used to design brand new viruses

Using artificial intelligence, researchers from Stanford University have created entirely new viruses that can replicate in laboratory settings. This marks the first instance of whole genomes being successfully designed by AI. The 16 novel viruses were engineered to target bacteria, posing no risk to human health. The achievement has been hailed as a "very significant turning point" in scientific progress, potentially paving the way for innovative treatments for diseases.

However, experts have cautioned that these AI-designed viruses raise significant safety and security concerns.

The research team utilized AI tools that function similarly to large language models, like ChatGPT, which predict sequences of text. In this case, the AI models, referred to as Evo1 and Evo2, predicted the genetic code of life rather than words. The models were trained on genetic data from a diverse range of organisms, including viruses, bacteria, plants, and humans. They were then fine-tuned to generate a specific type of virus called a bacteriophage, which can only infect certain species of bacteria.

Assistant Professor Brian Hie, who led the study, emphasized that this represents a new frontier in the complexity that can be designed using generative AI. The research process involved refining 302 AI-generated virus designs and subsequently synthesizing 16 of the most promising ones in the laboratory. Of these, 16 proved effective at killing E. coli bacteria.

The team's excitement upon seeing the phages working was palpable, with one researcher noting that the room spontaneously erupted in applause when the results were shared.

This breakthrough has the potential to revolutionize the treatment of antibiotic-resistant infections by enabling the development of new phage therapies. However, the study also highlights the dual nature of this technology, as it could be misused to create harmful biological agents. In a commentary published alongside the research in the journal Science, Dr. Thomas Inglesby and Dr. Moritz Hanke from the Center for Health Security at Johns Hopkins University emphasized the urgent need to address biosafety and biosecurity issues surrounding AI-designed viruses.

They stressed that it is no longer a matter of whether such technology will exist, but rather how it can be used responsibly to prevent serious harm.

To mitigate potential risks, the researchers took several safety precautions. They excluded viruses capable of infecting complex organisms from the training data, conducted the research on phages rather than human-infecting viruses, and carried out the experiments in a secure laboratory setting. Despite these safeguards, the study underscores the potential for AI to transcend the boundaries of natural biology and open up new possibilities in synthetic biology, which could lead to the development of novel drugs, therapies, and treatments for various diseases.