Artificial intelligence is transforming the design–build–test–learn cycle in synthetic biology. Instead of running years of iterative lab experiments, researchers now simulate and optimize designs in silico. First, tools like AlphaFold reshaped protein science by predicting complex folding patterns with remarkable accuracy. As a result, scientists began to trust AI models for structural biology. Soon after, genome-scale systems such as Evo 1 and Evo 2 expanded this capability from proteins to entire genomes.
In 2025, researchers pushed the boundary even further. They designed 16 synthetic bacteriophages using the blueprint of Phi X 174, which infects Escherichia coli. Scientists synthesized these viral genomes and tested them directly in the lab. Notably, several AI-designed phages infected bacteria more efficiently than natural strains. Therefore, AI no longer just predicts biological structures. It now engineers functional viral systems from the ground up.
“Artificial intelligence has moved from decoding life’s blueprint to actively drafting new versions of it.”
Fighting Superbugs and Cancer
Most importantly, AI-driven viral design offers powerful medical applications. As antimicrobial resistance rises, traditional antibiotics lose effectiveness. In contrast, AI-engineered bacteriophages target specific pathogens without harming beneficial microbes. For example, scientists can tailor phages to attack resistant bacteria such as Pseudomonas aeruginosa. Consequently, precision phage therapy could replace broad-spectrum drugs in high-risk infections.
Meanwhile, oncology research also benefits from this shift. Researchers now optimize oncolytic viruses to infect and destroy tumor cells selectively. In addition, AI refines viral payload delivery, which increases therapeutic precision. Vaccine development has advanced as well. By redesigning antigens computationally, scientists improve immune recognition against evolving pathogens like SARS-CoV-2. Furthermore, AI enhances guide RNA design in CRISPR-Cas9, reducing off-target edits and improving genome stability.
“Precision virology may soon outperform traditional drugs in both infection control and cancer therapy.”
The Dual-Use Dilemma and the Governance Imperative
However, this technological leap introduces serious biosecurity concerns. The same generative systems that design therapeutic phages could also enhance virulence or enable immune evasion. Moreover, advanced models can modify genetic sequences while preserving biological function. As a result, conventional DNA screening systems may fail to detect risky designs.
Although current experiments focus on bacteriophages that do not infect humans, the underlying methods apply broadly. Therefore, experts worry about accidental release, deliberate misuse, or regulatory gaps. In addition, cloud-based bioengineering platforms lower technical barriers. This accessibility accelerates innovation, yet it also increases exposure to dual-use risks.
To address these challenges, policymakers must strengthen DNA synthesis screening and implement function-based risk analysis. At the same time, developers should embed safeguards such as output monitoring and red-team testing within AI systems. Finally, international coordination remains essential. Without shared standards and transparent oversight, governance will lag behind innovation.
“Scientific progress demands speed but biosecurity demands vigilance.”
References
- King, S. H. et al. Generative design of novel bacteriophages with genome language models. Preprint at bioRxiv (2025).
- Mallapaty, S. World’s first AI-designed viruses a step towards AI-generated life. Nature (2025).
- Wittmann, B. J. et al. Strengthening nucleic acid biosecurity screening against generative protein design tools. Science (2025).
- National Academies of Sciences, Engineering, and Medicine. The Age of AI in the Life Sciences: Benefits and Biosecurity Considerations. Chapter 3: AI-Enabled Biological Design and the Risks of Synthetic Biology (2025).
- Hie, B. L. et al. (various works on protein and genome models referenced in Evo-related studies, e.g., Nature Biotechnology and related preprints, 2023–2025).
