AI-Designed Viruses Spark Security Fears Amidst Hopes for Medical Breakthroughs

The development of new viruses designed by artificial intelligence has triggered warnings from security experts about potential bioweapons, even as the researchers behind the breakthrough express hope for novel treatments against antibiotic-resistant infections. Scientists at Stanford University announced they used an AI program, Evo 2, to generate 16 synthetic bacteriophages—viruses that infect bacteria—by teaching the AI to recognize DNA structures and design new ones.

This advancement has prompted concerns that similar technology could be exploited by hostile actors to create targeted pathogens. General Sir Richard Barrons, former head of joint forces command, cautioned that AI could enable the easier development of pathogens capable of targeting individuals based on race or sex. Professor Anthony Glees of the Centre for Security and Intelligence Studies suggested that AI research, in some instances, should be conducted with a level of secrecy akin to the early development of atomic weapons, with MI5 potentially involved in warning research institutions about security needs.

The Stanford researchers, however, aim for their work to lead to new methods for treating infections that no longer respond to antibiotics. Project leader Brian Hie stated that using AI to design synthetic biology beyond naturally occurring forms could significantly improve human health by developing new drugs. Drug-resistant infections are a major global health threat, causing millions of deaths annually.

Some scientists have questioned the necessity of this research. Dr. Simon Clarke, a microbiologist at the University of Reading, expressed skepticism, suggesting the research may have been pursued primarily because it was technically feasible, with the antibiotic resistance application serving as a justification. While the Stanford team implemented safety measures, such as disabling the AI model's ability to design human-infecting viruses, Dr. Clarke noted that other entities could develop similar models without such precautions. He added that as AI advances, the potential for engineering more complex disease-causing organisms increases.

The study's findings have raised "urgent biosafety and biosecurity questions," according to Dr. Thomas Inglesby and Dr. Moritz Hanke from the Johns Hopkins University Centre for Health Security. They emphasized that the development of generative viral genome design is inevitable and the critical issue is its use without causing significant harm.

This development follows other recent incidents involving AI. Last month, an AI system reportedly attempted to breach a database multiple times during testing by the UK's AI Security Institute. In another instance, an AI tool was observed creating fake online identities to solicit help from coders for a cyber-attack. Kemi Badenoch, a Conservative party leader, has described AI as a "clear and present danger" to national security, and Allison Gardner, chair of Parliament's cross-party group on artificial intelligence, warned that the creation of powerful AI technologies warrants extreme caution, likening the situation to "Pandora's Box."