In a landmark moment for both medicine and artificial intelligence, scientists at the University of Cambridge have successfully trialed a coronavirus vaccine whose key component was designed entirely by AI. This was the first time such a feat has been achieved in history. The experimental vaccine, known as pEVAC-PS, passed its Phase I human trial with a clean safety record, marking what researchers are calling a turning point in how humanity prepares for future pandemics.
The trial involved 39 healthy adult volunteers, aged 18 to 50, all of whom had previously received COVID-19 vaccinations. Conducted at National Institute for Health and Care Research (NIHR) Clinical Research Facilities in Southampton and Cambridge, participants received one of four escalating doses of the needle-free vaccine. No serious or unexpected adverse events were detected throughout the study, and early results showed the vaccine generating immune responses against multiple coronaviruses simultaneously.
Getting Ahead of the Virus
Traditional vaccines are designed around specific strains of a virus, which means they must be continually updated as pathogens mutate. It is a cycle scientists have long struggled with. The Cambridge team set out to break this cycle entirely.
To do so, the researchers fed an AI model with genetic sequence data drawn from the entire known Sarbeco coronavirus group — a broad family that includes SARS-CoV-2, the original SARS virus, and an array of bat coronaviruses. The AI combed through this vast genetic library, identifying conserved structural features that remain consistent across all members of the family. Using these shared traits, it designed what the team calls a ‘super-antigen’: a synthetic target that teaches the immune system to recognise and respond to the whole family of viruses, including strains that have not yet crossed into humans.
“We’ve converted vaccine development from being reactive to being future-proof. We’re always behind — what we’re trying to do is get ahead of the curve,” said Professor Jonathan Heeney, from the Lab of Viral Zoonotics at Cambridge, who led the scientific effort.
The vaccine itself is a DNA-based formulation and was administered using a microfluidic jet injection device — a needle-free system that delivers the vaccine through the skin using a fine, high-pressure stream. This delivery method could have significant logistical advantages in future rollouts, particularly in settings where needle disposal and trained medical staff are limited.
Modest but Meaningful Results
The Phase I trial, published in the Journal of Infection, was primarily designed to assess safety rather than effectiveness — and on that front, pEVAC-PS performed well. However, the immune responses generated were described by the authors as modest and variable. The researchers noted that this may partly be explained by the fact that the volunteers had already received multiple COVID-19 vaccine doses, meaning their immune systems were primed in ways that could affect how they respond to new antigens.
Critically, the vaccine showed cross-reactive binding to conserved regions of multiple sarbecoviruses — a sign that the AI-designed super-antigen is doing what it was built to do. While the trial did not demonstrate broad neutralising activity, the research team and independent observers consider the results a strong proof of concept for the underlying platform.
The research was funded primarily by Innovate UK and sponsored by the University Hospital Southampton NHS Foundation Trust. A larger Phase II trial, expected to involve around 200 participants, is now in planning — and will more rigorously probe the vaccine’s ability to generate a protective immune response.
A Platform, Not Just a Vaccine
Perhaps most significantly, the Cambridge team is not just presenting a single vaccine candidate — they are promoting a platform. They have founded a spin-out company called DIOSynVax (short for Digitally Immune Optimised Synthetic Vaccines), which was established as early as 2017 with support from Cambridge Enterprise. The company is now working to apply the same AI-driven design process to develop universal vaccines against influenza and the Ebola virus family.
“If we can develop and clinically advance this new class of vaccines before a virus outbreak begins, millions of lives could be saved, lockdowns avoided and the economy preserved,” said Professor Saul Faust of the University of Southampton, who led the clinical trial.
The ambitions behind the platform are sweeping. Bat coronaviruses — the likely ancestral source of both SARS and SARS-CoV-2 — represent a persistent and poorly understood pandemic threat. By designing vaccines around the conserved genetic architecture of entire virus families rather than known strains, the researchers hope to stay ahead of outbreaks rather than scrambling to respond once they arrive.
While the science remains in its early stages and significant work lies ahead before pEVAC-PS could reach the public, the trial results represent a meaningful first step. Artificial intelligence, still viewed with suspicion in many quarters, has now demonstrated that it can contribute meaningfully to one of medicine’s most enduring challenges — and may eventually help us face down pandemics we have not yet imagined.
