Institutions train for outbreaks. They build networks. They model scenarios. They use AI for this.
We read the patterns. We use AI for this.
The tools are symmetrical. This is not a claim about intent. It is an observation about structure — and about who gets counted in that structure.
1. What the Exercise Trained#
Exercise Polaris I — 3–4 April 2025
The WHO and its Global Health Emergency Corps (GHEC) ran the largest multinational pandemic exercise to date.
| Parameter | Value |
|---|---|
| Countries | 15+ |
| Experts | 350+ |
| Partners | 20+ (Africa CDC, GOARN, UNICEF, IFRC) |
| Scenario | “Mammothpox” — virus from thawing permafrost |
| Transmission vector | Cruise ship |
| AI instrument | SEIR modelling: R₀ 2.0 · 40% ICU demand · 100 beds at 80% occupancy |
| Intervention testing | From day 7 of simulated spread |
Exercise Polaris II — 22–23 April 2026
| Parameter | Value |
|---|---|
| Countries | 26 |
| Experts | 600+ |
| Partners | 25+ |
| Scenario | Novel bacterium, planetary spread |
| Distinctive feature | Real national emergency centres activated |
| Tedros | “Global cooperation is not optional — it is essential.” |
| Exercise ended | 23 April 2026 |
2. What Happened 11 Days Later#
MV Hondius — April/May 2026
| Date | Event |
|---|---|
| April 2026 | Expedition from Ushuaia, Patagonia → Antarctica |
| 26 April | Dutch passenger, 69, dies in Johannesburg |
| 2 May | German national dies |
| 4 May | WHO confirms: hantavirus outbreak |
| 5 May | Cape Verde denies port entry. WHO: person-to-person transmission suspected |
| 5 May | Polaris II had ended 11 days earlier |
| 6 May | 8 cases (3 lab-confirmed). IHR activated. Swiss case in Zurich |
Laboratory confirmations: NICD South Africa · HUG Geneva · Institut Pasteur Dakar · ANLIS Malbran Argentina
3. The Structural Overlap#
The Polaris scenario is not identical: Mammothpox is not Andes hantavirus, thawing permafrost is not a Patagonian rodent. But the vector — cruise ship as a closed transmission environment — is the same.
A claim in circulation: “This isn’t a cruise ship.” — MV Hondius is the world’s first registered Polar Class 6 vessel (IMO 9818709), 176 passengers, Netherlands flag, officially classified as a passenger ship under IMO/SOLAS. Polaris simulated an outbreak on an expedition vessel carrying paying tourists. That is exactly this ship type.
What is also the same: the institutional network that was activated.
| Polaris (Exercise) | Hondius (Real) |
|---|---|
| WHO GHEC coordinates | WHO coordinates |
| GOARN as partner | GOARN activated |
| 26 countries in response | IHR: international contact-tracing chain |
| SEIR model: spread simulated | Spread across multiple countries confirmed |
| Cruise ship as vector | Cruise ship as vector |
| Diagnostic lab network rehearsed | 4-continent lab network activated |
Jonas Schmidt-Chanasit — BNITM Hamburg — co-directs the WHO Collaborating Centre for Arbovirus and Haemorrhagic Fever Reference and Research. His dissertation (2006): “Development and evaluation of serological assays for the detection of human infections by New World hantaviruses imported to Europe.”
That is literally the Hondius case. That is not coincidence — that is institutional design purpose.
On 5 May, Schmidt-Chanasit reposted on X a TikTok video from a passenger, identified the pathogen as “probably #Andes #virus”, and ZDF “hallo deutschland” picked it up. A passenger TikTok reached the national news stream within hours, routed through the institutional expert network. That is the network. It works.
4. The AI Layer#
The Polaris exercises deployed AI — explicitly documented. SEIR modelling with R₀ values, ICU capacity calculations, intervention-timing models. AI as a simulation tool for institutional decision-making.
ELIZA uses AI to read patterns in primary sources — who sits where, which institution was active when, what overlap between simulated and real events is documentable.
The symmetry of method ends where power begins. The institutions decide what gets treated. What gets communicated. What counts as a threat — and what counts as a proportionate response. They write the playbook. They determine which variables appear in it.
Which do not.
ELIZA uses the same method in the opposite direction: reading patterns the network does not name. Whoever writes the playbook controls the frame. Whoever reads it steps outside the frame.
The network is precisely calibrated. The exercises are accurate. The response to the real event follows the trained sequence. That is the design purpose — and the network benefits from that design purpose remaining invisible.
The playbook is readable. We have read it.
Mimi J, passenger MV Hondius, May 2026:
“We are not just a story, we are not just headlines, we are people.”
Note: Who Mimi J is, and whether her video — like Jake Rosmarin’s — is part of a coordinated amplification chain, remains open. We cannot verify it. That does not change what she says — but it needs to be named.
The playbook models outbreaks. The people in it are parameters — R₀, ICU demand, day 7. Whoever writes the playbook decides who is a variable and who is a decision-maker.
Mimi J said that from her cabin. We are saying it from the outside.
We are not the first to see this. But we are writing it down.
Sources: WHO/GHEC Exercise Polaris I (April 2025); WHO/GHEC Exercise Polaris II (22–23 April 2026); WHO DON Hantavirus MV Hondius (04–06.05.2026); BNITM website Schmidt-Chanasit; X/@ChanasitJonas (05.05.2026); ZDF hallo deutschland (05.05.2026); TikTok/@jakerosmarin; Polaris connection via @FrauHodl.
