On 24-25 September 2026, internet and phone networks went dark across Tigray as fighting resumed in northern Ethiopia. NetBlocks said its data showed a subnational internet disruption in northern Ethiopia that matched reports of a telecoms blackout in Tigray. Reuters reported that calls to Ethio Telecom numbers did not connect and that Safaricom customers had only limited connectivity.
Who ordered the cut is disputed. Reuters, citing four people with knowledge of the matter including a diplomatic source, said the TPLF told both Ethio Telecom and Safaricom's Ethiopian subsidiary to shut the networks. A source close to the TPLF, speaking anonymously, denied responsibility and blamed the government. Witnesses quoted by Ethiopia Observer said Ethio Telecom itself cut access. None of these claims is confirmed on the record. The companies and the Prime Minister's office did not respond to Reuters. Local reports say service returned in Mekelle around 4 October, with much of the region still disrupted around 10 October. I found no reliable report of the current status.
The strongest case for a cut
It is worth stating the security argument fairly. Armed conflict can make telecom networks a military asset. One source told Reuters the aim could be to neutralise government drones, and a risk advisory seen by Reuters said transformers had been removed in several towns. A government could argue that denying an adversary communications saves lives. A rebel authority could argue the same about the state's targeting. If the reports of a TPLF order are right, the logic is the same in both directions. Every party to a conflict has a tactical reason to want the network off.
Why that case does not hold up
The security rationale runs into three problems that the evidence on Tigray already shows.
First, the costs fall on civilians, not combatants. The earlier Tigray shutdown began in November 2020 and lasted nearly two years. Context reported it affected about six million people, and almost half of them were in severe need of food. The African Union estimates that the 2020-2022 war caused around 600,000 deaths from combat, famine and the collapse of healthcare. Communications blackouts do not cause those deaths alone, but they blind families, aid agencies and clinics to what is happening. Al Jazeera notes that the earlier blackout and blockade produced famine-like conditions for hundreds of thousands of people.
Second, a shutdown is an economic and trust shock well beyond the target area. Context cites NetBlocks estimating that Ethiopia's 23-day nationwide outage in June 2020 cost more than $100 million. For a country trying to build a digital economy, each blackout teaches investors, developers and users that the network can disappear on a political decision.
Third, a blackout hides the facts. Both sides in this dispute are making claims about who cut access. Without independent reporting, nobody outside can check claims about airports seized, casualties or forced recruitment. Reuters said it could not independently confirm the Ethiopian military's casualty figures. Opacity helps whoever has the most to hide.
What the law says
The legal standard is clearer than the politics. The African Commission on Human and Peoples' Rights adopted its Resolution on Internet Shutdowns and Elections in Africa (ACHPR/Res.580) in 2024. It calls on states to refrain from ordering the interruption of telecommunications services or shutting down the internet. That resolution concerns elections and is not binding by itself, so it does not squarely cover a wartime blackout. But it shows where the continent's main human rights body stands, and it applies to a state that has ratified the African Charter, as Ethiopia has.
The sharper precedent is judicial. In Amnesty International Togo and 7 Others v. Republic of Togo (ECW/CCJ/JUD/09/20), the ECOWAS Court of Justice ruled in 2020 that Togo's 2017 internet shutdown was unlawful. Secondary summaries say the Court relied on the absence of a legal framework authorising it. That is the key point for proportionate regulation. Even if a state can show a real emergency, it must act under a published law that limits who can order a cut, for how long, and with what review. A secret instruction to a carrier meets none of those tests.
The Togo ruling also shows the limits of enforcement. Reporting a year later said Togo had not complied with the Court's orders. Ethiopia is outside ECOWAS in any case, so the case is persuasive authority there rather than binding.
What proportionate policy looks like
The policy lessons are practical.
- Treat operators as a constraint on abuse. If networks go dark on an unwritten phone call, a carrier cannot refuse. Licences and emergency rules should require written, time-limited orders that operators can publish or challenge. This applies to state-owned carriers and to private ones such as Safaricom.
- Make attribution possible. Operators should publish outage notices and share technical data with independent monitors such as NetBlocks, so claims about who cut access can be tested.
- Protect humanitarian channels. Even in a real conflict, hospitals, aid agencies and the ICRC need exempt connectivity. The ICRC had facilitated some 116,000 phone calls and oral messages during the earlier Tigray blackout, which shows how large the gap is when networks fail.
- Press every party, not just the state. The Reuters report, if accurate, shows that armed non-state actors can also order networks off. Digital rights advocacy should not make an exception for them.
The pro-innovation position is not that governments may never act in a crisis. It is that blanket shutdowns are blunt, costly and rarely effective, and that targeted, lawful, reviewable measures are better. Until Ethiopia and its regional institutions say who ordered this cut and on what authority, Tigray's civilians are again the ones paying.