An AI Designed a Universal Coronavirus Vaccine - and It Just Passed Its First Human Trial

For the first time, a vaccine whose active ingredient was designed by artificial intelligence has been tested in people - and it did what its makers hoped. A team from the University of Cambridge and the University of Southampton, working with the Cambridge spin-out DIOSynVax, let machine learning read the shared genetic code of an entire family of coronaviruses and design a single synthetic target from the features they all have in common. In a first-in-human trial, the resulting needle-free DNA vaccine was safe and triggered broad immune responses. It is an early result - but a real proof of concept for a future-proof way of making vaccines.

1. The problem: vaccines have always been reactive

For most of their history, vaccines have been built one threat at a time. A new virus or a new variant appears, scientists isolate it, and they design a shot aimed at that specific target. It works - but it is a race that always starts a step behind the virus. The 2020s made the cost of that lag vivid, as updated shots chased one variant after another.

The question the Cambridge-Southampton team set out to answer was simple to state and hard to do: instead of designing a vaccine against the virus in front of us, could we design one against a whole family of viruses at once - including the members that have not emerged yet?

2. How an AI designed the vaccine

The target family is the sarbecoviruses - a subgroup of coronaviruses that includes the original SARS virus (SARS-CoV-1), the COVID-19 virus (SARS-CoV-2), and a wide range of related coronaviruses circulating in bats, some with the potential to jump to humans. The members differ in many ways, but they also share deep similarities - conserved features that the virus cannot easily change without breaking itself.

DIOSynVax’s approach was to turn that biology into a computing problem. Machine learning analysed genetic-sequence data from across the sarbecovirus group worldwide, mapped the structures they have in common, and used that to engineer a single synthetic super-antigen: a designed molecule that does not copy any one virus, but presents the immune system with the shared, hard-to-mutate features of the entire family. Train the body on the common core, the logic goes, and the protection should be broad - and pointed at threats that do not exist yet.

3. The trial and what it showed

The first-in-human study enrolled 39 healthy adults aged 18 to 50 at NIHR clinical research facilities in Southampton and Cambridge. The results were what a first trial is meant to deliver.

“We’ve converted vaccine development from being reactive to being future-proof.”
- Professor Jonathan Heeney, University of Cambridge, founder of DIOSynVax

4. The needle-free twist

The vaccine is delivered as a DNA vaccine: rather than injecting a ready-made protein, it carries genetic instructions so the body’s own cells briefly produce the designed target and learn to recognise it. In this trial it was administered without a needle, using a microfluidic jet that pushes a fine stream of fluid through the skin. DNA vaccines are typically stable and quick to reprogram for a new target - useful traits if the aim is a platform that can be updated and deployed fast.

5. Why it matters

Two threads make this notable. The first is the design philosophy: a broadly protective, family-wide vaccine would shift preparedness from chasing each new virus to being ready before one arrives. The second is the method: if a computer can design an antigen that behaves correctly in real human immune systems, it points toward faster, cheaper, more systematic vaccine design - and a template that could be aimed at other dangerous virus families, from influenza to filoviruses.

It is also a quietly encouraging story about what AI is for. The same tools that make headlines elsewhere here did something concrete and hopeful: read a mountain of viral genetics, find the common thread, and hand scientists a candidate that passed its first human test.

What we still don’t know (the honest caveats)

• This is a Phase 1, first-in-human trial in 39 people. It shows the vaccine is safe and prompts the right kind of immune response - not that it prevents infection or illness in the real world.
• Demonstrating real-world protection requires larger Phase 2 and efficacy trials, in more diverse populations and over longer follow-up.
• A strong immune response in the lab does not always translate into strong real-world protection; that link still has to be proven for this vaccine.
• The approach is a platform and a proof of concept - promising, and years of work from a deployable product.

Sources

• First-in-human trial paper, Journal of Infection (June 2026). DOI 10.1016/j.jinf.2026.106759
• University of Cambridge: New ‘universal vaccine’ technology could protect us from future virus outbreaks
• MedicalXpress: AI-designed universal vaccine clears first human trial · ScienceDaily: AI-designed universal coronavirus vaccine passes first human trial

Curated by Jerry Cards - jerrycards.com. We read the week’s most consequential science, health and tech research so you don’t have to. More at jerrycards.com/news.