AI-designed vaccine passes its first human trial
A universal vaccine against sarbecovirus coronaviruses, whose active component was conceived by artificial intelligence, has safely cleared a first-phase trial in 39 volunteers in Britain.
- A universal vaccine against sarbecovirus coronaviruses, whose active component was entirely conceived by **artificial intelligence**, was developed by the **University of Cambridge** and the company **DIOSynVax**.
- The first-phase trial involved **39 healthy volunteers** aged 18 to 50, at centres in Southampton and Cambridge.
- The vaccine was judged safe, with no significant adverse effects, and triggered an immune response to SARS-CoV-2, SARS and related bat coronaviruses.
- The results were published in the journal **Journal of Infection** in June 2026, with a larger phase two now planned.
The University of Cambridge and the biotechnology company DIOSynVax have reported that a universal vaccine against the sarbecovirus group of coronaviruses, whose active component was entirely conceived by artificial intelligence, has successfully cleared a first-phase clinical trial in humans. According to the available data, it is the first time a vaccine designed wholly through computer simulations has been tested in people.
The research is led by Professor Jonathan Heeney of the Department of Veterinary Medicine in Cambridge. Rather than the classic approach, in which a vaccine is built around part of a single known virus, the AI here analysed global genetic sequences across the entire sarbecovirus group and devised a synthetic super-antigen carrying structural features shared by that family of viruses. The aim is for such a component to teach the immune system to recognise not only existing strains but also related viruses that might yet jump from animals to humans.
How did artificial intelligence design the vaccine?
The first-phase trial was led by Professor Saul Faust as chief investigator and enrolled 39 healthy volunteers aged 18 to 50. It was carried out at research centres in Southampton and Cambridge. A phase-one study is not designed to prove effectiveness; it checks whether a vaccine is safe and whether it provokes any immune response at all.
According to the published results, the vaccine was judged safe, with no significant adverse effects. Participants showed an immune response aimed at several related viruses at once:
- SARS-CoV-2, the virus that causes covid-19
- SARS, which caused an outbreak in the early 2000s
- related bat coronaviruses from the same sarbecovirus group
That response profile is what the authors describe as universality: a single vaccine targeting a structure shared by an entire family of viruses, rather than one strain alone. The results were published in the peer-reviewed journal Journal of Infection in June 2026.
What this result does not prove
It is worth separating what has been shown from what still needs confirming. A phase-one study in 39 people shows that the candidate is safe and prompts an immune response, but it says nothing reliable about how well it protects against actual infection, how long that protection lasts, or how it behaves in older people and patients with chronic illness, who were not covered by this 18-to-50 age band.
Professor Heeney described the significance of the approach this way:
We have turned vaccine development from reactive into future-proof.
To test that claim, a larger phase-two trial is planned, taking in more participants and wider age groups. Only the later phases, if successful, could show whether a computer-designed component really offers broad and durable protection against future strains.
The wider message of this work is not one particular vaccine but a method: if the active part can be reliably designed through simulation, defences against the next coronavirus could begin before a new strain even appears. Should computer-designed vaccines pass the same rigorous trial phases as conventional ones before entering wider use?
