Researchers at Stanford University have successfully used Artificial Intelligence (AI) to design the first fully functional bacteriophages—viruses that infect bacteria—capable of replicating in a laboratory setting. In a study published in Science, the team demonstrated that AI models Evo1 and Evo2 could generate whole viral genomes from scratch, a milestone described as a "very significant turning point" in scientific research.
The breakthrough marks the first time generative AI has been used to design a complete, viable genome that can replicate and function within cells. The researchers trained the AI models on two million bacteriophage genomes, refining the technology to produce viruses targeting specific bacterial species. From 302 AI-generated designs, the team synthesized and tested the viruses in the lab, with 16 proving effective at killing E. coli bacteria.
Core findings of the study:
- The AI-generated bacteriophages pose no threat to humans, animals, or plants, as they are engineered to infect only bacteria.
- The technology functions similarly to large language models like ChatGPT, but instead of predicting text, it predicts genetic sequences.
- The research was conducted under controlled laboratory conditions, with no release of the engineered viruses into the environment.
Reactions and concerns
The study has drawn both praise and criticism, highlighting the dual-use nature of AI in biotechnology. Supporters emphasize the potential for new treatments, including targeted antibacterial therapies and advancements in synthetic biology. Critics, however, warn of urgent biosafety and biosecurity risks, citing the lack of governance frameworks to regulate such technology.
In an accompanying commentary published in Science, Dr. Thomas Inglesby and Dr. Maurice Hanke from Johns Hopkins University stressed that while the research holds promise, it also raises serious questions about oversight. They wrote: "Although this is promising for life sciences applications, it also raises urgent biosafety and biosecurity questions. The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not."
How the AI models work
The AI tools Evo1 and Evo2 were trained on a vast dataset of bacteriophage genomes, enabling them to predict and generate new genetic sequences. Unlike traditional genetic engineering, which relies on modifying existing organisms, this approach allows for the de novo design of viruses—a process previously considered highly complex.
Dr. Brian Hie, the lead researcher and a chemical engineer at Stanford, described the work as "new territory" for generative AI in biology. "In this case, we wanted the model to generate the entire genome end-to-end in a single left-to-right pass," he said. "We didn’t add anything." The team’s goal was to demonstrate that AI could autonomously design functional biological systems, a capability with broad implications for medicine and biotechnology.
Potential applications
Researchers suggest the technology could accelerate the development of:
- Novel antibacterial treatments, particularly against drug-resistant bacteria.
- Customized bacteriophages for targeted bacterial infections in agriculture or medicine.
- Advanced synthetic biology tools, including AI-designed organisms for industrial or environmental applications.
Safety measures and ethical considerations
The study was conducted in a high-containment laboratory, and the researchers emphasized that the engineered viruses were not released into the environment. However, the research has reignited debates about the responsible use of AI in biology, with calls for international guidelines to prevent misuse.
Next steps for the research
The Stanford team plans to expand testing to other bacterial species and refine the AI models for greater precision. They also aim to collaborate with regulatory bodies and ethicists to address concerns about biosecurity and unintended consequences.
The breakthrough underscores the rapid pace of AI advancements in science while highlighting the need for proactive governance to ensure these tools are used safely and ethically.