Researchers at the University of Cambridge have completed the first-ever human clinical trial of a vaccine component designed entirely by artificial intelligence. The Phase 1 trial, involving 39 healthy volunteers, demonstrated that the AI-crafted antigen was safe and capable of generating immune responses against not just SARS-CoV-2, but also SARS and bat-borne coronaviruses that haven’t yet jumped to humans.
How an algorithm built a better antigen
The project was led by Prof. Jonathan Heeney of the Cambridge Lab of Viral Zoonotics and DIOSynVax (DVX) Ltd, a spinout company focused on computationally designed vaccines. Their AI system ingests genetic sequence data from global coronavirus surveillance, essentially reading the molecular blueprints of every known Sarbecovirus strain circulating in animals and humans.
From that data, the machine learning model identifies conserved features, the structural elements that remain stable across different strains even as viruses mutate. Instead of targeting the parts of a virus that change rapidly, the AI zeroes in on the parts that stay the same.
The output is what the team calls a “super-antigen,” a single synthetic protein designed to train the immune system to recognize a broad family of coronaviruses rather than just one specific variant.
The 39 volunteers in the trial were healthy adults aged 18 to 50. Trial data was collected between December 2021 and September 2023, and the results were published in the Journal of Infection. No significant side effects were reported, and the vaccine successfully triggered immune responses against multiple Sarbeco coronaviruses.
Why this matters beyond the lab
This is the first time a computationally designed vaccine component has been tested in humans. Every previous AI-assisted drug or vaccine project has used machine learning as a tool within a human-led design process. This trial represents a case where the AI system itself architected the core immunological component.
Prof. Heeney’s team is essentially trying to build pandemic insurance, protection against viruses that haven’t spilled over into humans yet but are circulating in bat populations and could emerge at any time. Given that SARS-CoV-1 appeared in 2002, MERS in 2012, and SARS-CoV-2 in 2019, the pattern suggests another Sarbecovirus spillover isn’t a matter of if but when.
What this means for the AI and biotech landscape
For the intersection of AI and healthcare, this trial is a proof of concept that will be watched closely. The pharmaceutical industry has been pouring billions into AI-driven drug discovery, with companies like Insilico Medicine, Recursion Pharmaceuticals, and others racing to bring AI-designed molecules through clinical trials. Cambridge’s result gives the entire field a credibility boost, even though this is just Phase 1 data and larger efficacy trials will be needed before any product reaches the public.