A Phase 1 clinical trial in Southampton has produced the first published evidence that a vaccine designed entirely by artificial intelligence can safely trigger immune responses in humans. The results, appearing in the Journal of Infection, mark a milestone for a novel approach to vaccine development that aims to prepare for future pandemics before they emerge.
The vaccine candidate was created by researchers at the University of Cambridge and their spinout company DIOSynVax. Rather than modifying an existing virus protein, machine-learning algorithms analyzed genetic data from every known Sarbeco coronavirus. The AI identified shared molecular features across the viral family and stitched them into a single synthetic antigen designed to train the immune system against stable, conserved parts of the virus structure. Traditional vaccines target the spike protein of a specific variant, meaning mutations can render them less effective—a pattern that defined much of the COVID-19 pandemic response.
The AI-designed antigen takes a different approach by focusing on parts of the virus that do not change, teaching the immune system to recognize the structural core of an entire family of threats. The trial enrolled healthy volunteers in Southampton and Cambridge.
No significant side effects were reported, meeting the primary safety requirement for a Phase 1 study. Immune data showed that participants developed antibodies capable of neutralizing multiple coronaviruses: SARS-CoV-2, the original SARS virus that killed 774 people in 2003, and bat coronaviruses that have never infected humans but are considered potential spillover risks. The vaccine was delivered using a needle-free jet injector, which applies a high-pressure stream of fluid to the skin.
This method eliminates the fear of needles that deters some people from vaccination and removes logistical requirements for syringe supply and sharps disposal. Researchers describe the injector as simpler, faster, and cheaper to deploy at scale.
The Cambridge approach represents a fundamental shift in vaccine development logic. The current paradigm is reactive: a virus emerges, scientists sequence it, design a vaccine, manufacture it, and distribute it—a process that takes months. The AI-powered platform flips this sequence by analyzing viral families in advance and designing antigens for threats that do not yet exist.
Researchers say the same platform could be adapted for Ebola, influenza, or any rapidly mutating virus. As a Phase 1 trial, the study was small and designed to test safety and immune response rather than real-world protection against infection or disease.
Larger studies will be required to confirm that the immune response translates into actual prevention. The results, however, provide the first human data point for an AI-designed antigen passing initial safety and immunogenicity hurdles. The next pandemic virus is already circulating in animal reservoirs such as bats, birds, or pigs.
The trial suggests a potential path forward: using AI to map viral families in advance, design universal antigens, and have a safe vaccine candidate ready to test before the first human case is ever reported.
