Cambridge’s AI-designed bird flu vaccine protects mice and ferrets across variants

Mice and ferrets given an AI-designed bird flu vaccine were fully protected against several genetically distinct variants, a Cambridge–DIOSynVax study reported. Only about half the animals given the stockpiled comparator vaccine survived exposure to the same viruses. The result supports a broader approach to pandemic preparation, but human trials are still the next step.

The vaccine was developed by the University of Cambridge and its spinoff, DIOSynVax. The study, published in Cell Reports Medicine, examined protection against bird flu variants that have caused human infections and represent potential pandemic threats. Cambridge described the findings in a November 5 account published by Medical Xpress.

The researchers are addressing a moving target. Bird flu is an RNA virus, a group that can mutate rapidly into new strains. Those changes can help viruses evade immune responses and existing vaccines. The team’s aim is to design protection around features shared by many variants, rather than follow each newly detected strain.

From surveillance sequences to a vaccine candidate

The design process began with genetic sequences collected through bird flu surveillance programs around the world. Those programs sample viruses from infected wildlife, livestock and humans. The researchers used an AI tool to integrate the sequence data and design candidate vaccine ingredients containing antigen features common across the viruses.

An antigen is the ingredient that prompts the immune system to mount a protective response. It trains the body to recognize a pathogen carrying that antigen during a later infection. The team calls its engineered candidates “super-antigens” because they combine features shared across the group, rather than simply targeting one existing variant.

AI produced a range of candidates, not a finished vaccine ready for use. Researchers screened those candidates in cells, selected the best one and then tested the vaccine in animals. Jonathan Heeney, the Cambridge professor who led the work, said AI greatly accelerates design; laboratory testing then checks whether the candidates are robust and safe.

It's designed to anticipate viral evolution, rather than reacting to it,

Jonathan Heeney, University of Cambridge, describing the experimental vaccine

A strain-specific stockpile meets a broader test

The comparator reflects the current approach to pandemic preparedness: governments stockpile vaccines against virus variants that have already emerged. The bird flu vaccine used for comparison was based on representative virus isolates identified through the World Health Organization’s pandemic influenza preparedness program. It was designed to protect against one specific strain.

That creates a mismatch risk as viruses change. Cambridge’s account explains that traditional strain-specific vaccines can offer limited protection by the time they are manufactured and distributed, requiring regular updates. In this experiment, exposing vaccinated animals to genetically distinct viruses tested protection beyond a single strain—the breadth the new design is intended to deliver.

Human protection remains the next hurdle

The team intends the approach to cover existing variants and potentially related viruses that have not yet emerged in humans. That is a design goal, not something this animal experiment can establish for every future virus. The reported result concerns the genetically distinct bird flu variants used in the study.

The researchers’ next step is human clinical trials, with the aim of having a vaccine ready when needed. The present study does not establish protection in people. Its contribution is an animal-tested candidate and a direct comparison supporting further development—not a bird flu shot already demonstrated to work in humans.