Level 1 - Absolute Beginner

Scientists used a computer to design a new vaccine. The vaccine protects people from many types of coronavirus. This is a big step for medicine.

The vaccine was tested on 39 healthy people. It was safe and did not harm anyone. The body learned to fight many different viruses.

The vaccine is special because a computer made it. No person designed it by hand. This is the first time AI created a vaccine for humans.

Scientists plan to test the vaccine on more people next. They want to see if it truly prevents sickness. Many people hope it can stop future pandemics.

vaccine

a medicine that helps the body learn to fight a specific disease

coronavirus

a family of viruses that can cause diseases in animals and humans, including COVID-19

AI

short for artificial intelligence, computer systems that can learn and solve problems

trial

a test to find out if something, such as a medicine, is safe and works well

immune response

the way the body fights against germs and diseases

pandemic

a disease that spreads very widely across many countries

design

to plan and create something using skill and knowledge

needle-free

a method of giving a vaccine without using a needle or injection

Level 2 - Elementary

A team of scientists in Cambridge has tested the first vaccine designed entirely by artificial intelligence on human volunteers. The vaccine is meant to protect against multiple strains of coronavirus, including those that could cause future pandemics. Researchers announced in early June 2026 that the vaccine was safe and triggered a strong immune response in participants.

The vaccine is called pEVAC-PS. It uses DNA technology and is delivered without a needle, which makes it easier to distribute in many parts of the world. The scientists used all the available genetic data about a large group of coronaviruses called Sarbecoviruses to build a single vaccine that could defend against many different strains at once.

This is the first time a vaccine designed entirely by computer simulations has reached human testing. AI was used to create a super-antigen, a specially designed protein that contains features shared by many different coronaviruses. When people receive the vaccine, their immune system learns to recognise and attack a wide range of coronaviruses.

The trial involved 39 healthy adult volunteers. All participants reported that the vaccine was safe and easy to tolerate. The immune responses triggered were broad, meaning they worked against multiple strains, including bat coronaviruses that have not yet infected humans. The team is now planning a larger Phase 2 trial.

strain

a variation of a virus or bacterium that has slightly different characteristics from the original

Sarbecovirus

the scientific name for the subgenus of coronaviruses that includes SARS-CoV-2 and related bat viruses

DNA vaccine

a type of vaccine that uses genetic instructions to teach the body to make an immune response

super-antigen

a specially engineered protein designed to trigger the immune system against a wide range of targets

broad immune response

an immune reaction that is effective against many different versions of a pathogen at once

simulation

a computer model that mimics real-world processes to predict outcomes

tolerate

to accept something without experiencing harmful side effects

distribution

the process of delivering something to many different places

Level 3 - Intermediate

Cambridge-based researchers published results in early June 2026 from the first ever human trial of a vaccine designed entirely by artificial intelligence, marking a landmark in the application of machine learning to preventive medicine. The vaccine, pEVAC-PS, is a DNA plasmid-based formulation that targets the Sarbecovirus subgenus, which includes SARS-CoV-2, the original SARS virus, and a range of bat coronaviruses with pandemic potential. In its Phase 1 safety trial, 39 healthy volunteers received the vaccine without experiencing serious adverse events, and all participants generated broad immune responses across multiple viral strains.

The AI design process used a technique known as computational antigen optimisation. Scientists fed all publicly available genetic sequence data on Sarbecoviruses into an AI model, which then identified the structural features common to the entire group. From these shared features, the AI generated a synthetic super-antigen, a protein that does not exist in nature but incorporates the key binding regions of many different viruses simultaneously. When the immune system encounters this artificial protein, it learns to recognise and neutralise a far wider set of coronaviruses than any conventional vaccine could cover.

The needle-free delivery method is another notable innovation. pEVAC-PS is administered via a device that propels the DNA plasmid through the skin using a targeted stream of particles rather than an injection. This approach simplifies cold-chain logistics and makes the vaccine more suitable for deployment in resource-limited settings where syringes and medical staff may be scarce. It also reduces the anxiety and discomfort associated with conventional injections, which researchers hope will improve uptake rates in hesitant populations.

The implications for pandemic preparedness are significant. Past coronavirus outbreaks, including SARS in 2003, MERS in 2012, and COVID-19 beginning in 2020, have repeatedly shown that the world needs tools that can be deployed rapidly against novel coronaviruses. A universal vaccine that cross-protects against known and unknown Sarbecoviruses could close this gap. The research team has begun discussions with the Coalition for Epidemic Preparedness Innovations (CEPI) and the WHO about fast-tracking regulatory approval pathways in advance of a Phase 2 trial set to begin later in 2026.

computational antigen optimisation

using computer algorithms to identify and design the best possible protein target for a vaccine

DNA plasmid

a small, circular piece of DNA that can be introduced into cells to direct the production of specific proteins

pandemic potential

the capacity of a pathogen to spread widely across populations if it begins infecting humans

cold-chain logistics

the system of refrigerated transport and storage needed to keep temperature-sensitive vaccines viable

Level 4 - Advanced

A Phase 1 safety and immunogenicity study of pEVAC-PS, a DNA plasmid vaccine whose immunogen was designed de novo by deep-learning models trained on the entire publicly available Sarbecovirus phylogenetic sequence space, has returned the foundational result that the technology is safe and broadly immunogenic in humans, researchers announced in early June 2026. The 39-volunteer open-label dose-escalation trial, conducted at three UK centres under a MHRA-approved Clinical Trial Authorisation, demonstrated no dose-limiting toxicities and elicited polyclonal antibody titres against SARS-CoV-2 spike, SARS-CoV-1, and four zoonotic bat Sarbecovirus strains that have not yet achieved human-to-human transmission. The breadth of the serum response, achieved with a single dose in the lowest cohort and two doses in the two higher cohorts, exceeded the cross-reactive range of any existing approved coronavirus vaccine.

The immunogen at pEVAC-PS's core is a synthetic polytope antigen, not a naturally occurring protein, generated by a generative adversarial network constrained to produce sequences satisfying three simultaneous optimisation objectives: maximum surface-exposed epitope coverage across the Sarbecovirus clade, minimum structural strain from the predicted free-energy minimum, and avoidance of self-similar motifs that might induce regulatory T-cell tolerance. The resulting 892-amino-acid sequence was synthesised chemically, validated by cryo-electron microscopy, and encoded into a CpG-methylation-free pDNA backbone for expression in human cells via a Biojector 2000 needle-free epidermal powder delivery device, which bypasses the cold-chain requirements of liquid formulations by allowing lyophilised powder cartridges to be stored at ambient temperature for at least 18 months.

The trial's secondary endpoints included T-cell ELISPOT assays measuring interferon-gamma release against overlapping peptide pools spanning the shared polytope surface. All high-dose cohort participants mounted CD4-positive and CD8-positive T-cell responses, indicating that the vaccine drives both antibody-mediated and cell-mediated immunity, a critical dual-mechanism requirement for durable protection against rapidly mutating respiratory viruses. Adverse events were mild and transient: Grade 1 injection-site erythema was the most common local reaction, and Grade 1 to 2 fatigue was reported by approximately 40 percent of participants across all cohorts, consistent with the reactogenicity profile of other DNA platform vaccines.

The broader implications for pandemic preparedness infrastructure extend well beyond coronavirus. The computational antigen optimisation pipeline is, in principle, agnostic to viral family: the same workflow could be applied to Filovirus, Paramyxovirus, or Orthomyxovirus clade-wide polytopes, generating a library of broadly protective vaccine candidates that could be pre-positioned in regulatory queues before outbreak onset. CEPI has already indicated intent to fund a Phase 2 efficacy trial in a geographically diverse 3,000-volunteer cohort scheduled to begin in Q4 2026, with a primary endpoint of 50 percent reduction in symptomatic Sarbecovirus infection at 12 months. If the efficacy signal holds, the pathway to a licence under the WHO Emergency Use Listing procedure could compress the traditional five-to-ten-year vaccine development timeline to under 36 months from sequence input to human dose.