voidly
Active Incident

Confirmed: sustained censorship in Ethiopia (2026-03)

Ethiopia · March 19, 2026 · 10 days

OONI detected 21 days of elevated network interference averaging 66.5% anomaly rate.

Evidence

66%
Anomaly Rate
21
Measurements
77%
Confidence

What This Means

66% of 21 network tests from Ethiopia failed to reach these services. Global baseline failure rate is typically 1-3%. This 63+ percentage point difference strongly indicates intentional blocking.

How We Detected This

Data Sourceooni
ML ModelGradientBoosting v3.3 (LOCO median F1 0.87, honest across 127 countries)
Confidence Score77% certainty of censorship
Detection TimeMarch 19, 2026

Corroborating Evidence (21)

Evidence collected from multiple independent sources. Upstream claims shown in quotes are not Voidly assertions.

OONIOpen Observatory of Network Interference21 records
middlebox-detection
"OONI http_invalid_request_line: 3/5 measurements anomalous (60.0%)"OONI report
Signal value: 0.60
Mar 29, 2026
View source →
header-manipulation
"OONI http_header_field_manipulation: 3/5 measurements anomalous (60.0%)"OONI report
Signal value: 0.60
Mar 29, 2026
View source →
blocking
"OONI whatsapp: 6/7 measurements anomalous (85.7%)"OONI report
Signal value: 0.86
Mar 27, 2026
View source →
blocking
"OONI whatsapp: 1/2 measurements anomalous (50.0%)"OONI report
Signal value: 0.50
Mar 27, 2026
View source →
blocking
"OONI whatsapp: 3/6 measurements anomalous (50.0%)"OONI report
Signal value: 0.50
Mar 27, 2026
View source →
+ 16 more OONI records

Affected Services (2)

TorWhatsApp

What we cannot say

  • Who caused it — measurement data alone does not attribute this to any government, ISP, or other actor.
  • Intent — anomalous or unreachable measurements indicate interference, not a stated motive or a deliberate-vs-incidental determination.
  • Which networks — no specific ASN/ISP is identified for this incident.
  • The mechanism — we cannot state whether this was DNS tampering, TLS/SNI reset, an HTTP blockpage, throttling, or an infrastructure outage.
  • Sub-national scope — measurement is aggregated at the country level; we cannot localize to a province, city, or network segment.
  • Independent corroboration — this rests on a single source; treat it as a single-source signal, not a multi-source confirmation.

Cite This Incident

Voidly Research. (2026). Confirmed: sustained censorship in Ethiopia (2026-03) [ET-2026-0098]. Voidly Global Censorship Index. https://voidly.ai/incident/ET-2026-0098

Related

Related incidents