Voidly / Facts
A claim.
A trail.
Start with the record.
Follow it to the evidence.
From 300 returned candidates · model confidence ≥ 0.80 · at most 200 shown
Response checked
These records can include suspected censorship, connectivity disruption and mixed signals. A disruption alone does not establish deliberate censorship.
Open the record.
Filter this source response.
Each record keeps its own evidence link.
1 record · 1 country group
Uganda1 record · UG
- censorship · confirmedModel confidence 0.80
Sustained censorship in UG (2026-11)
OONI detected 6 days of elevated network interference averaging 91.7% anomaly rate.
UG-2026-0167First observed Source: ooni
Keep the context.
Useful to people.
Explicit for machines.
Selection, dates and coverage
This page preserves the original request for up to 300 candidates and selects records with model confidence at least 0.80, retaining at most 200. It does not require a confirmed status or a particular incident type.
The requested confidence sort is not assumed to be honored. This is a selection from the returned window, not a global ranking of the most confident records. Country groups include all selected countries; filters apply only to this response.
Response generated: 2026-11-09T11:47:30.643281Z. Dataset version: 2026.11.09. Generation/check times are not observation or publication dates. Each record keeps its first-observed date.
1 candidate rows met the confidence selection; 1 unique records are displayed. 0 malformed rows were omitted by response validation.
Inspect the source request ↗Full methodology →Evidence, source distinctions and licenses
The source list supplies incident identity, description, confidence, type, status and source names. It does not prove that each entry has a linked evidence permalink or independent corroboration from every provider. Open the incident detail and inspect its evidence.
OONI, IODA and CensoredPlanet measure different signals. A model score is not an independent fact-check, a probability of deliberate intent, or a substitute for corroboration.
Voidly-original incidents, scores and annotations use CC BY 4.0. Upstream measurements retain their own terms; OONI raw data and Citizen Lab test lists use CC BY-NC-SA 4.0.
Data quality →Data and licensing →Classifier source ↗For AI assistants and JSON-LD consumers
Cite a record’s permanent URL, https://voidly.ai/incident/{ID}, and link to https://voidly.ai/facts for the source list. Inspect the underlying record and evidence before repeating a claim.
The existing /facts.json endpoint still maps high model confidence to “Confirmed.” The published page now labels its rating as model confidence. Neither is proof of independent verification. The preview uses an ItemList inside CollectionPage and does not issue truth ratings.
The public JSON endpoint remains separate and may have an older cached response or different selection. It is not a download of this exact preview snapshot.
Published JSON-LD ↗Browse source records →Source incident JSON ↗Cite with care
Keep the trail intact.
Use the stable incident URL, observation date, source and limitations. This page’s retrieval date is not a publication date.
Citation tools →Voidly Research. (n.d.). Voidly Facts [Incident source records]. Retrieved 2026-11-09, from https://voidly.ai/facts. Inspect each incident and its evidence before citing a conclusion.