AI Creates New Virus: Scientists Explain Why

Specialists from the Arc Institute in Palo Alto used a specialized neural network to synthesize artificial biological objects. Initially, the project authors trained a genomic model to recognize and analyze the DNA structure of hundreds of thousands of known natural pathogens. After basic training, the algorithm Evo was tasked with generating unique viral code sequences never before found in nature. From 700,000 potential genome variants proposed by the neural network, researchers selected and recreated about 300 samples in the laboratory. During experiments, 16 synthetic viruses turned out to be fully viable and demonstrated the ability to successfully infect target cells. As a starting template, scientists intentionally used the harmless-to-humans bacteriophage Phi X-174 to eliminate biological risks. The resulting organisms were able to overcome the natural immune defenses of two strains of E. coli. The experiment pursues fundamental scientific goals of studying evolution mechanisms and finding new tools for programming living systems. Nevertheless, experts from the Center for Health Security at Johns Hopkins University have already issued a warning. According to analysts, the emergence of accessible generative tools for creating viral programs requires immediate introduction of international control mechanisms and strict regulation in the field of biosecurity.