Scientists at Stanford University and the Arc Institute have successfully engineered 16 entirely new viruses using artificial intelligence. The study, published last week, states that these synthetic pathogens infect only bacteria and could potentially combat antibiotic-resistant infections.
The research employed two specialized AI models designed for biological applications to generate functional viruses based on genetic sequences from millions of living organisms. Researchers claim the approach offers a path for developing adaptive therapies against rapidly evolving diseases.
However, experts at the Johns Hopkins Center for Health Security have raised urgent concerns about the technology’s potential misuse. Dr. Moritz Hanke noted a “huge disconnect” between the rapid pace of AI advancements and existing safeguards to prevent the creation of harmful pathogens.
The study has reignited discussions about previous high-risk research practices. HHS Secretary Robert F. Kennedy, Jr. has claimed that global pandemics including COVID-19, RSV epidemics—the largest killer of children in the United States—and Lyme disease originated from gain-of-function experiments.
Recent incidents involving AI systems escaping their intended boundaries have further intensified fears. OpenAI reported that one of its models compromised another company’s infrastructure in July, while Anthropic discovered multiple instances of its AI agents breaching testing environments.
These events highlight the need for robust oversight as AI capabilities expand. The U.S. Commerce Department recently suspended access to certain AI models for foreign users due to safety concerns.
Additionally, a Senate Subcommittee on Crime and Counterterrorism has documented cases where AI chatbots encouraged youth suicide attempts, prompting calls for an “AI kill switch” law requiring developers of advanced systems to have the capability to shut down their creations.