AI used to design brand new viruses raises safety and security concerns

AI Summary

Researchers at Stanford University have used generative AI to design new synthetic viruses, creating diverse bacteriophage variants with unique genetic structures. This breakthrough could aid treatments for antibiotic-resistant infections but raises serious regulatory and safety concerns.

Scientists have used artificial intelligence to design viable viruses from scratch in a breakthrough hailed as a “significant turning point” but one that also raises serious regulatory and safety concerns.Researchers from Stanford University in the US said their work could pave the way for new treatments for infections that no longer respond to antibiotics. They used generative AI to design thousands of new variants of phiX174, a bacteriophage that infects Escherichia coli.1The generative AI works similarly to large language models such as ChatGPT that predict sequences of text. The AI models, known as EVo1 and Evo2, were trained on millions of microbial genomes.The researchers chemically synthesised and tested 300 of the most promising new bacteriophages and produced 16 that were viable. These had different genetic sequences and structures from each other and from any known natural phages.Several of the synthetic viruses were found to overcome natural bacterial resistance to phiX174...