Pandemic preparedness
The world still cannot reliably stop a new pandemic virus before it spreads widely.
Open in the interactive tree →COVID-19 showed that vaccines can be made within a year, but not in time and not fairly shared. Preparedness means early detection, rapid vaccines for any virus family, equitable access and political will.
As of October 2026
The WHO Pandemic Agreement (adopted May 2025) still lacks its annex on sharing pathogen samples and benefits and cannot yet be opened for signature; details are in the node on the pandemic agreement. H5N1 bird flu has infected 993 people worldwide since 2003 with 477 deaths (WHO figures to January 2026), and has spread through US dairy cattle since March 2024, although sustained human-to-human spread has not been seen. Moderna started phase 3 trials of an mRNA H5N1 vaccine candidate in April 2026.
What is missing
- Finished treaty rules (the pathogen-access and benefit-sharing annex) and ratification by 60 countries
- Global early-warning surveillance, genomic and wastewater, in every region
- Broadly protective vaccines against whole virus families, ready before an outbreak
- Manufacturing and distribution capacity in every region, financed in advance
- Trust and cooperation between governments
Becomes possible once solved
- Vaccines within 100 days of a new threat
- Universal flu and coronavirus vaccines
- Outbreaks stopped locally
Open steps
- Earlier outbreak detection Medium AI leverageDetect a new pathogen from wastewater, clinical, genomic and news signals before it spreads widely, in every region.
- Predicting immune escape High AI leveragePredict which virus mutations escape immunity so that vaccines and antibodies can be chosen ahead of time.
- Vaccines for whole virus families Medium AI leverageDesign antigens and antibodies that protect against a whole virus family and are ready before an outbreak.
- Spillover risk Medium AI leveragePredict which animal viruses can infect humans and where contact is most likely, to aim surveillance.
- Response planning models Low AI leverageModel outbreak spread, interventions and supply chains so that responses can be chosen fast with honest uncertainty.
Where AI could help
Medium AI leverage. AI helps predict immune escape and speeds early warning and vaccine design; treaty politics, funding, factories and trust remain binding.
- Scanning news, flight, wastewater and genomic data for outbreak signals
- Predicting which viral mutations escape immunity and pre-testing vaccine designs
- Designing antibodies and broadly protective antigens for whole virus families
- Modeling spread and supply chains for response planning
Shown so far
- In October 2023 Nature published EVEscape, a deep-learning tool trained on pre-2020 sequences that anticipated SARS-CoV-2 escape mutations about as well as lab scans and also works for influenza, HIV, Lassa and Nipah. source
- In November 2025 Nature published RFdiffusion-designed antibodies whose binding poses were confirmed by cryo-EM against targets including influenza hemagglutinin. source
- According to the company, BlueDot's AI sent customers an alert about the Wuhan pneumonia cluster on 31 December 2019, days before the WHO public notice. source
Prerequisites
- Quarantine1377Isolating arrivals is the oldest tool for slowing an outbreak
- Vaccination1796
- Epidemiology1854
- World Health Organization1948WHO runs global outbreak alerts and the International Health Regulations
- Genome sequencing2003
- COVID-19 pandemic2020COVID-19 showed vaccines can come within a year, but not in time or fairly shared
- mRNA technology2020
- WHO Pandemic Agreement2025
Unlocks
- Infections under control2040s?