The Anomaly
On July 22-23, 2026, independent network-measurement platforms IODA (Internet Outage Detection and Analysis, run by Georgia Tech's Internet Intelligence Lab) and OONI (Open Observatory of Network Interference) recorded a nationwide connectivity anomaly in Ethiopia — a 7% disruption rate against a normal 1-3% baseline, spread across Addis Ababa, Amhara, Oromiya and the Southern Nations, Nationalities and Peoples' Region. It is not (yet) the kind of full mobile-data blackout that shut Amhara off from the internet for eleven months between August 2023 and July 2024. But the timing is not coincidental. It lands amid intensifying fighting between the Fano militia and federal forces, and weeks after Ethiopia's national election produced some of the worst documented civilian violence of the current conflict cycle.
The Amhara Association of America's July 10, 2026 monitoring statement recorded 62 human-rights-violation incidents across 43 districts and 15 zonal administrations in Amhara and Oromia during June alone — 495 civilians affected, including 112 killed, 134 injured and three cases of sexual violence. Human Rights Watch's World Report 2026 separately documents drone strikes, abductions of teachers, and the killing of an aid worker in the same conflict. This is the backdrop against which a national network anomaly should be read.
The Case for the Government's Position
It's worth stating plainly what Addis Ababa's argument would be, because it isn't baseless. Fano has used encrypted platforms to coordinate ambushes and, per HRW's reporting, to organize abductions of teachers and civil servants. Security services in active-conflict zones genuinely do lose real-time visibility into troop movements and civilian evacuation routes when adversaries use commercial networks freely. Governments fighting a real insurgency — and Ethiopia's is real, with thousands of documented battle events in Amhara this year — have a legitimate interest in denying an armed group a coordination tool. That is a materially different claim than "shutdowns stop violence," and it deserves to be engaged with rather than dismissed as pretext.
Why the Justification Collapses in Practice
The problem is that Ethiopia's own recent history is the best evidence against this argument. The 2023-24 Amhara blackout — nearly a year, covering a region home to roughly a quarter of the country's population — did not reduce the human-rights violations UN investigators and the Ethiopian Human Rights Commission documented over that period; it just made them harder to verify from outside. Fano's offensive capacity was, by the government's own later claims of hundreds of fighters killed in 2025, undiminished by a blackout that had already ended. What a shutdown reliably degrades is not insurgent coordination — Fano still operates in the field — but civilians' ability to reach banking, remittances, medical referrals and family members, and journalists' and rights monitors' ability to document what is happening to them in real time. A tool marketed as counterinsurgency functions, in effect, as an accountability blackout.
The Structural Problem
This is possible at all because Ethiopia's connectivity runs through a single chokepoint. Ethio Telecom holds effective control of the country's telecommunications backbone, and, as Freedom House's Freedom on the Net assessment documents, there is no transparent legal procedure specifying when or why the state cuts access — no published emergency criteria, no judicial sign-off, no notice requirement. A state of emergency proclamation, invoked at will, has functioned as sufficient legal cover. That is a policy choice, not a technical necessity: countries with competitive, multi-carrier internet markets find shutdowns far costlier and harder to execute cleanly, because no single entity controls enough of the market to flip the switch.
The Regional Accountability Gap
Africa has a stated continental norm against exactly this behavior. The African Commission on Human and Peoples' Rights adopted Resolution 580 in March 2024, explicitly urging member states to preserve unrestricted internet access before, during and after elections. Ethiopia held a national election in June 2026. Yet according to CIPESA's tracking of KeepItOn coalition data, continent-wide shutdown incidents rose from 17 in 2023 to 21 in 2024 — after the resolution passed, not before. A soft-law resolution with no enforcement mechanism and no cost attached to violating it will not outcompete a government's short-term security calculus, however imperfect that calculus is.
What Should Follow
The honest position is not that Ethiopia's security concerns are fabricated, but that shutdowns are a poor instrument for addressing them and a good one for suppressing scrutiny of civilian harm — and the two effects are hard to disentangle after the fact, which is precisely the problem. A government confident that a partial network anomaly reflects legitimate operational security, rather than the opening move of a broader blackout, should have no objection to publishing the legal basis for the disruption and a defined end condition. Regional bodies serious about enforcing Resolution 580 need a mechanism with teeth — trade, AU-summit, or telecom-licensing consequences — not another statement. Until then, expect the anomaly graphs from IODA and OONI to keep flagging Ethiopia first.