AI-Designed 'Super Antigen' Vaccine for Coronaviruses Enters Phase 2 Trial After Proving Safe in Humans

In a Phase 1 clinical trial involving 39 healthy volunteers, a needle-free jet injector delivered a synthetic DNA sequence that does not exist anywhere in nature. The candidate, known as pEVAC-PS, is the world's first viral antigen designed entirely by artificial intelligence to be tested in human subjects. Developed by researchers at the University of Cambridge and its spinout company DIOSynVax, the vaccine has successfully cleared its initial safety hurdles, marking a critical milestone for computational immunology. By demonstrating that an AI-generated genetic sequence can be safely administered to humans without triggering severe adverse events, the trial opens the door to a new era of proactive pandemic preparedness, and the candidate is now advancing toward larger Phase 2 efficacy trials.[1][4]

The results of the trial, recently published in the peer-reviewed Journal of Infection, represent a potential paradigm shift in how the global medical community defends against emerging infectious diseases. Rather than waiting for a novel pathogen to spill over into the human population and then rushing to sequence and target it, researchers are using machine learning to build preemptive defenses against entire families of viruses at once. The dose-escalation study administered the vaccine at four different strength levels, monitoring participants closely for reactogenicity. The data confirmed that the AI-designed vaccine was well-tolerated across all cohorts, with no significant safety concerns or debilitating side effects reported, providing the foundational safety data required to push the experimental technology forward.[2]

Traditional vaccine development is an inherently reactive process that leaves populations vulnerable to viral evolution. Current immunizations, including the highly successful mRNA platforms, train the body's immune system to recognize a specific, circulating strain of a virus. However, as pathogens like SARS-CoV-2 replicate and spread, they frequently mutate, altering their surface spike proteins to evade this highly specific trained immunity. This evolutionary arms race forces scientists into a constant, expensive cycle of updating formulations and rolling out seasonal boosters. This "strain-chasing" approach means that by the time a new vaccine is manufactured, distributed, and administered, the virus has often already mutated again, leaving public health officials perpetually one step behind the pathogen.[1][5]

To break this reactive cycle, the Cambridge research team deployed artificial intelligence to analyze vast, global genomic databases of the sarbecovirus subgenus. This viral family includes the original SARS virus that emerged in 2003, SARS-CoV-2, and numerous related bat coronaviruses that scientists have identified as carrying a high risk for future zoonotic spillover. The AI was tasked with looking past the rapidly mutating surface proteins to identify conserved structural features that are shared across the entire viral family. These conserved regions represent the essential biological architecture that a virus cannot easily mutate or discard without losing its fundamental ability to function, replicate, and infect host cells.[1][3]

Using these universally shared, unchangeable features, the AI computed a single "super-antigen"—a mosaic protein blueprint designed to train the human immune system to recognize the foundational structure of the virus family, rather than the specific disguise of a single variant. Because this synthetic antigen targets the unchangeable core of the pathogen, the resulting immunity is theoretically "future-proofed." It is designed to provide lasting protection against both the current mutations circulating in the population and related viruses that currently reside only in animal reservoirs but could trigger the next global health crisis if they make the jump to humans.[4]

The Phase 1 trial provided the first tangible clinical evidence supporting this ambitious computational approach. Blood samples collected from the 39 volunteers revealed that the pEVAC-PS vaccine successfully triggered cross-reactive immune responses. The antibodies generated by the participants' immune systems were capable of recognizing and binding to not only SARS-CoV-2 and the original SARS virus, but also several distinct bat coronaviruses that have never actually infected humans. While laboratory assays are not a guarantee of real-world protection, this cross-reactivity suggests that the AI-designed vaccine could indeed offer a preemptive, broad-spectrum shield against future zoonotic spillover events before they have the chance to escalate into pandemics.[2][4]

Beyond its novel AI-driven design, the pEVAC-PS vaccine also introduces significant logistical advantages that could transform global health equity. The candidate is formulated as a DNA-based vaccine, which is inherently much more thermostable than the fragile mRNA vaccines that dominated the COVID-19 response. This high degree of stability eliminates the need for the complex, ultra-cold storage chains that severely hampered vaccine distribution in developing nations. By remaining stable at standard refrigeration or even room temperatures, the DNA vaccine is vastly easier to stockpile for years and transport to remote, resource-limited regions where pandemic outbreaks frequently originate.[2][5]

Furthermore, the vaccine is administered intradermally using a specialized PharmaJet Tropis device—a microfluidic system that utilizes a high-pressure fluid stream to deliver the dose precisely through the skin without the use of a traditional needle. This needle-free approach offers a welcome alternative for the significant portion of the population that experiences injection anxiety, which can be a barrier to vaccine uptake. More importantly for public health logistics, the jet injector system streamlines the administration process, reducing medical waste and allowing healthcare workers to conduct faster, safer mass vaccination campaigns during an active and rapidly spreading outbreak.[2]

Despite the highly promising safety profile and innovative delivery mechanisms, infectious disease experts caution that the technology is still in its earliest clinical stages. The Phase 1 trial was explicitly designed to assess safety and tolerability, not real-world effectiveness against active infections. While the vaccine did successfully generate an immune response, some independent analyses of the data characterized the resulting antibody levels as modest when compared to the highly targeted, strain-specific vaccines currently on the market. The true test of the platform will be whether this broad, cross-reactive immunity translates into robust, lasting clinical protection against severe disease when patients are exposed to live viruses.[3]

To answer those critical efficacy questions, researchers are now actively preparing for a much larger Phase 2 trial. This next stage will enroll a significantly broader and more diverse population to better evaluate the vaccine's immunogenicity across different age groups, demographics, and complex immune histories. If the super-antigen concept proves effective at scale, the implications for modern medicine extend far beyond coronaviruses. The same AI-driven platform is already being adapted to develop universal vaccine candidates for other high-risk pathogen families, including influenza and Ebola, potentially giving humanity the tools to disarm future pandemics before they ever begin.[1][4]