A technician at a Siberian lab has died of yersinia infection aka plague. The details are fuzzy, but the lab seems to have been working with pneumonic plague, an especially contagious and dangerous form. A few articles quote authorities as saying the risk is small.
What about forecasters?
Sentinel Global Risks Watch1 are the people best placed to have an opinion on this. They are a team of tournament-winning superforecasters who provide probabilities on upcoming risks as a public service, and their track record is excellent. They say:
Our forecasters think there’s a 0.7% (0.1% to 2.0%) chance that there will be more than 1,000 deaths attributed to plague across the globe in the next 12 months. The low probability stems from the lack of known further cases, the initial rapid isolation of contacts, and the ability to contain the spread of typical plague outbreaks with antibiotics. However, we note the possibility that the worker may have died from infection with an antibiotic-resistant strain of plague or from a completely different disease that may or may not have been related to her work at the institute. While some commentators online are drumming up alarm, we mostly think this is unjustified at this time because there is currently no report of onward spread of any disease and strict quarantine measures have reportedly been implemented. Thus, we consider this a serious incident that merits monitoring but not alarm at this time. It is also a good opportunity for readers to consider taking some general pandemic preparedness measures in their own homes, buy some PPE, perhaps some far-UVC lights2, think about what would be good rural places to go to and how, etc., not because of this incident in particular but taking it as a prompt.
Some of our forecasters think that the real story is, to some extent, not whether there was a pneumonic plague death or whether there might be a small pneumonic plague outbreak, which happen periodically around the world, but whether and the extent to which the Russian government might have been developing bioweapons in the worker’s institute, and whether the current situation may have resulted from a bioweapons lab accident. This incident is also worrying because it confirms our estimation that the risks from lab leaks, and especially bioweapons lab leaks, are a latent worry.
What about the AI superforecasters? AIs recently beat the top human forecasters in the Metaculus Cup for the first time, so the machines may now be even more trustworthy than Sentinel. I asked two.
Probability: 1%. The baseline rate of global plague mortality is far below the 1,000-death threshold, typically averaging around 60 deaths per year. According to the WHO’s 2019–2025 review, 10 countries reported 3,847 suspected cases and 423 deaths over the seven-year period . The highest single-year death toll during this time was 79 in 2023 . Crucially, these counts already incorporate suspected, probable, and confirmed cases across all clinical forms—bubonic, pneumonic, and septicaemic . Exceeding 1,000 deaths would require an event roughly 13 times deadlier than the worst recent year.
The recent death of a lab worker at a Russian anti-plague institute in Irkutsk in early October 2026 raises vigilance, but current evidence suggests the incident is contained. Russian authorities deny a lab accident and report “pneumonia of unknown aetiology” . The WHO states that Russia has reported no confirmed plague cases and assesses the risk of broader spread as very low . Approximately 190 contacts have been identified, monitored, and tested with no dangerous pathogens detected or secondary symptomatic cases emerging . Because pneumonic plague develops rapidly, the lack of onward transmission weeks after the initial exposure strongly indicates this chain of infection is over, leaving the probability of an outbreak exceeding 1,000 deaths highly remote.
On further questioning, it said that even that 1% was mostly just that the regular epidemics in Madagascar got especially bad this year, and the chance of a global pandemic starting with the Russian lab was “effectively zero”.
0.2% chance. I assign less than 1% probability to a YES resolution. The Russian incident has not established secondary cases or sustained transmission (ECDC assessment). YES requires both an exceptional mortality surge and qualifying WHO reporting that places enough deaths inside the specified window.
What about prediction markets? Polymarket is currently at 5% for WHO declaring plague a “pandemic”, but markets are notoriously bad at handling very small probabilities, so I don’t take it seriously as a counter to the sources above.
I think this is a good example of a situation where technical forecasting is helpful. I felt underserved by mainstream articles, which either tried to maintain journalistic neutrality by giving no opinion, or quoted scientists’ vague statements saying not to worry. Given that all three top forecaster sources were 1% or less, and the statement they were forecasting was much weaker than the scenario we’re worried about (eg it includes other sources of plague epidemics, it starts at 1000 deaths rather than a true global pandemic, etc), I think they’re sending a strong message that this isn’t a direct cause for concern (although I agree with Sentinel that this is a good indirect reminder that lab leaks, bioweapons, and pandemics all remain ever-present dangers).
Conflict of interest note: An ACX grantee
Conflict of interest note: That link goes to Aerolamp, another ACX grantee.