An AI-designed vaccine targeting the entire sarbecovirus family, the coronavirus subgroup responsible for both the 2003 SARS outbreak and COVID-19, has passed its first human clinical trial, researchers at the University of Cambridge have announced.
The vaccine, pEVAC-PS, was developed by Cambridge spin-out DIOSynVax using an AI platform that analysed global genetic sequence data across all known sarbecoviruses and identified the conserved molecular features shared by the entire family. Rather than targeting a specific strain, it encodes a synthetic “super-antigen” representing a common factor of a virus group and is explicitly designed to remain effective even as new variants emerge.
In a phase I dose-escalation trial involving 39 healthy volunteers at NIHR facilities in Cambridge and Southampton, pEVAC-PS was safe and well-tolerated at all doses. The team detected cross-reactive antibody responses against multiple sarbecoviruses, including SARS-CoV-2, SARS, and related bat coronaviruses.
There are caveats: the trial was designed for safety rather than efficacy, neutralising antibody responses were not robust, and pre-existing coronavirus immunity in most participants complicates interpretation. A larger Phase II trial is planned. The same AI platform is being developed for influenza and haemorrhagic fever viruses, including Ebola, fostering hope for future vaccines.
There are further advantages over current vaccination systems. The delivery system used involves a needle-free microfluidic jet and the thermostable DNA platform requires no ultra-cold storage. The DNA vaccine’s thermostability and needle-free delivery are genuine advantages for low-resource settings. But pEVAC-PS is a commercialised spin-out product developed with UK public funding. Without explicit access commitments now, history gives us every reason to ask whether the populations most vulnerable to pandemic coronaviruses will benefit, or whether this will follow the familiar pattern of innovation for affluent markets first.
Full trial data: Munro et al., Journal of Infection, 2026