AI developer Anthropic has announced that its Claude models are supporting local health authorities and international organizations responding to an outbreak of Bundibugyo virus (BDBV), a rare strain of Ebola, in the Democratic Republic of the Congo (DRC). Claude is being used to help compile case counts, organize laboratory data, trace transmission chains, and compare vaccine candidates, among other tasks.
A response effort buried in a day of manual tallying
The BDBV outbreak was confirmed in the DRC in May 2026, with an emergency declared on May 17 of that year. The virus has since spread across seven provinces and 63 of the country's 167 health zones, with confirmed cases approaching 8,000. The fatality rate stands at around 48 percent, making early detection critical: how quickly a sick person is found and isolated, before symptoms turn severe, often determines whether they survive. In North Kivu, treatment center bed occupancy has reached 94.7 percent.
Case information typically originates with community health workers going door to door, is relayed to health facilities and districts over WhatsApp, compiled into PowerPoint decks, and then passed up to the outbreak coordination center and provincial health ministries. Each handoff in that chain adds delay before decision-makers see an accurate picture of the outbreak. Anthropic, working with the Coalition for Epidemic Preparedness Innovations (CEPI), the WHO Regional Office for Africa, and the DRC's National Institute of Biomedical Research (INRB), has built Claude into this reporting pipeline. Claude reads the case and lab figures out of each district's PowerPoint slides, compares them against the previous day's numbers, flags meaningful changes, and summarizes district-level updates. As a result, what used to be a full day of situation report preparation now takes under an hour.
Four ways Claude is being put to work
Beyond automating situation reports, Anthropic describes three additional roles Claude is playing in the response. The first is disease modeling: teams that previously had time to run only a single forecasting model can now run several in parallel, giving them more evidence to decide where to position treatment centers. The second is organizing vaccine development data, where Claude has built a dashboard that tracks incoming proposals and supports comparisons such as cross-reactivity assessments across candidates. The third is genomic analysis through Claude Science, which assembles viral genomes from fragmentary sequencing data and builds phylogenetic trees to help trace how new infections spread and identify emerging variants. Claude is also being used to clean up "line lists" for a concurrent chikungunya outbreak in the DRC.
Humans still make the calls
Anthropic emphasizes that Claude functions purely as decision support, and that interpreting its output and deciding how to respond remains the job of human experts. Dr. Jean-Jacques Muyembe, a DRC public health expert involved in the response, has said that beating an outbreak like this one means knowing where the virus is today, not where it was a week ago. Paul Ouma of the WHO Regional Office for Africa noted that teams can now run several models where time once allowed only one. At the same time, faster report preparation does not by itself prove better health outcomes, and no data has yet been published showing that the forecasting models themselves have become more accurate.
Summary
Anthropic's Claude is now supporting the DRC's Ebola response across four fronts: automating situation reports, running disease models, organizing vaccine development data, and analyzing viral genomes, cutting a task that used to take a full day down to under an hour. The case offers a concrete look at how far AI can go in an active outbreak response, though whether faster reporting translates into slower spread or better treatment outcomes remains something to watch for going forward.
