Keep Up with Addis Insight
Add us to your Google Preferred Sources to see updates first.
From Tigray to Amhara and Oromia, the expanding use of unmanned strikes is leaving a growing record of reported deaths, damaged civilian infrastructure and unanswered questions about accountability.
ADDIS INSIGHT | IN-DEPTH ANALYSIS | 16 NOVEMBER 2026
At least 62 reported deaths in three incidents over nine days
On October 29, 2026, strikes hit the Jari Number 2 displacement camp in South Wollo, Amhara. Wire-service accounts placed the toll at roughly 14–16 dead, while subsequent BBC Amharic reporting cited 18 deaths, more than 30 injured, and four to five strikes spaced approximately 15–20 minutes apart. The reported victims included pregnant women and students. The accounts differ and should not be treated as a single independently established total.
On October 30, civilians collecting grain spilled from a previously struck truck near Rama in Tigray were hit. Reports cited 11 hospital-confirmed deaths, with other accounts giving 12 or 13. On November 6, five witnesses interviewed by BBC Amharic and BBC Tigrinya reported that two strikes hit a group of 200–300 displaced people traveling at night through Wedesa Dima in Kiremu woreda, Oromia, killing at least 36 and injuring at least 16. The BBC could not independently verify that incident, and the responsible operator was not established.
Using the lower reported totals of 14, 11 and 36 gives at least 61 deaths across these three reported incidents; using 18, 12 and 36 yields 66. Neither is a verified nationwide tally. The cases nevertheless show how quickly reported civilian losses can accumulate in separate theaters.
The numbers behind the crisis
| Indicator | Reported figure | Important qualification |
| Jari Number 2, 29 Oct 2026 | 14–18 killed; 22 to more than 30 wounded | Different outlets reported different totals |
| Rama, 30 Oct 2026 | 11–13 killed | 11 hospital-confirmed; other witness/media counts differ |
| Kiremu, 6 Nov 2026 | At least 36 killed; at least 16 wounded | Witness-based; not independently verified by BBC |
| Kiremu group size | 200–300 displaced people | Eyewitness estimate |
| Alamata, 22 Sep 2026 | 14 killed; 22 injured | Civilian/combatant breakdown unknown |
| Aid truck, Sep 2026 | More than 41 metric tonnes of food | Cargo quantity, not casualties |
| Explosive weapons, 2013–Feb 2024 | 2,273 civilians killed or injured | All explosive weapons, NOT drones alone |
| Health-care attacks, Jan 2025–Apr 2026 | 89 incidents | All weapons and perpetrators; NOT drone-only |
Source basis: the supplied research compilation, which attributes individual figures to BBC Amharic/Tigrinya, AFP, Anadolu, ACLED, Human Rights Watch, Action on Armed Violence and humanitarian monitors. Numbers are presented as reported, not independently audited.
How Ethiopia’s drone campaign spread
The technology’s military importance rose during the 2020–2022 Tigray war. As the battlefield shifted in 2021, the federal military expanded access to Turkish, Chinese and Iranian unmanned aircraft. Drone surveillance and guided strikes offered a way to monitor roads and target vehicles without exposing pilots to ground fire. By late 2021, imagery had documented Bayraktar TB2 aircraft at Bahir Dar. The shift changed the battlefield, but it also created a new danger for civilians in places where fighters, roads, shelters and food distribution routes overlapped.
The January 7, 2022 strike on a school compound sheltering displaced people at Dedebit became a defining case. At least 57 people were reported killed and more than 42 injured. Human Rights Watch identified fragments consistent with Turkish MAM-L guided munitions and reported no evidence of a military objective at the site. On January 10, a strike near May Tsebri reportedly killed 17 people at a grain mill, most of them women.
After the November 2022 Pretoria agreement, the conflict geography changed rather than the drone capability disappearing. Operations against armed groups in Amhara and Oromia brought aerial attacks into a more dispersed environment: market towns, roads, agricultural areas and residential settlements. In August 2023, a strike in Finote Selam reportedly killed 26 civilians. In February 2024, an attack on a passenger vehicle near Sasit reportedly killed at least 30 and injured 18. The precise circumstances and responsibility for individual incidents require case-specific scrutiny.
A chronology of reported civilian harm
| Date | Place | Reported outcome |
| 7 Jan 2022 | Dedebit, Tigray | At least 57 killed; more than 42 injured |
| 10 Jan 2022 | May Tsebri area, Tigray | 17 killed at grain mill, according to reports |
| 19 Oct 2022 | Metta Welkite, Oromia | Reports of follow-on strike affecting rescuers; count disputed |
| Aug 2023 | Finote Selam, Amhara | 26 civilians reportedly killed |
| Feb 2024 | Sasit area, Amhara | At least 30 killed; 18 injured in reported vehicle strike |
| 22 Sep 2026 | Alamata, Tigray | 14 killed; 22 injured; status of victims unclear |
| 29 Oct 2026 | Jari Number 2, Amhara | 14–18 reported killed |
| 30 Oct 2026 | Rama, Tigray | 11–13 reported killed |
| 6 Nov 2026 | Kiremu, Oromia | At least 36 reported killed; independent confirmation lacking |
The problem with counting drone deaths
There is no independently audited, nationwide total that cleanly separates civilian casualties from drone strikes, manned-aircraft attacks, artillery and other explosive weapons. This is not a minor technical problem: in several incidents, witnesses and organizations disagree over the platform used. The supplied research notes that Human Rights Watch described some 2022 Tigray attacks as ‘airstrikes’ without specifying whether drones or conventional aircraft were responsible.
Action on Armed Violence counted 2,273 civilians killed or injured by explosive weapons of all types in Ethiopia between 2013 and February 2024. Of these, 1,965 were harmed from 2021 onward and 1,027 were recorded as killed across the full period. Those numbers establish the wider burden of explosive violence, not a drone-only death toll. Presenting them as drone casualties would materially mislead readers.
Similarly, Insecurity Insight’s 89 recorded attacks on health care from January 2025 to April 2026 are not all drone attacks. A survey of 602 health workers covering abuses from November 2020 through September 2024 found 63.5% reporting attacks on ambulances, 61.1% theft or looting, 57% armed takeovers of facilities, 45% attacks on colleagues and 28% direct targeting. These survey percentages describe respondents’ experiences across multiple armed actors and types of violence; they cannot be converted into drone strike totals.
What the 2026 incidents reveal
Displacement camps are not safe simply because they shelter civilians
The Jari Number 2 accounts describe repeated impacts on temporary shelters occupied by families who had already fled earlier insecurity. Reporting cited 36 casualties brought to Hayq Primary Hospital, with 12 serious cases transferred onward to Dessie. Those hospital figures must not be added to separate injury estimates as though they represented additional victims: they may overlap.
Follow-on attacks can make rescue and food collection deadly
The Rama case is especially revealing because the reported victims were gathering spilled grain from a truck struck the previous evening. A separate 2022 account from Metta Welkite described students going out to help the wounded after an initial strike before another attack occurred. These accounts raise questions about whether people returning to an impact site were recognized as civilians and whether operators took feasible precautions before further attacks.
Night movement becomes a civilian risk
Witnesses in Kiremu described a group of 200–300 displaced people walking by flashlight at around 10 p.m. The reported dead included eight children younger than five, 17 women, including five pregnant women, three blind monks and three elderly men. These categories may overlap; they should not be summed into an invented demographic total. The group reportedly suffered two consecutive strikes, and survivors described fear of further attack while attempting to bury the dead or assist the wounded.
The limits of ‘precision’
Precision-guided weapons can reduce the physical distance between a selected aim point and an impact point. They cannot by themselves ensure that the selected person, vehicle or building is a lawful target. A strike can be technically accurate and still cause unlawful civilian harm if the intelligence is wrong, if precautions are inadequate or if expected civilian losses are excessive relative to the concrete and direct military advantage anticipated.
International humanitarian law requires distinction between civilians and combatants, proportionality and feasible precautions. A school or hospital is not automatically a lawful target because fighters are alleged to be nearby. Conversely, the presence of military personnel in a civilian site can change the legal analysis, depending on the circumstances. Publicly available casualty figures alone cannot establish criminal intent or the legality of every individual strike.
Why the strategy persists
Drones offer long surveillance endurance, lower exposure for pilots and access to roads and valleys that are difficult for ground forces to control. The supplied research cites rough purchase-price comparisons of around $5 million for a TB2 and $1 million–$2 million for some Wing Loong systems, against far higher costs for many conventional fighter aircraft. Such figures are approximate platform estimates, not audited Ethiopian procurement costs or complete operating expenses.
The appeal is particularly strong in dispersed counterinsurgency campaigns. Yet remote strikes cannot settle the political disputes sustaining conflict. Aerial superiority may restrict armed movement while simultaneously deepening civilian fear and distrust, especially when explanations or credible investigations do not follow civilian casualties.
Accountability and what remains unknown
For several recent incidents, the supplied research says military and regional officials did not answer journalists’ requests for comment. It also describes communications interruptions, restrictions on access and the end of a UN investigative mandate as barriers to comprehensive verification. These conditions can delay identification of victims, weapon types, military objectives and the actors responsible.
The appropriate public-interest questions are specific: Who authorized each strike? What information identified the target? What civilian presence was known or reasonably foreseeable? Were warnings or other precautions feasible? Were post-strike assessments conducted? Have families been informed of findings or offered remedies? The absence of public answers is not proof of every allegation, but it makes scrutiny much harder.
The human cost is larger than the casualty tables
Aerial surveillance can change behavior without a weapon being fired. Families may avoid markets, medical journeys, funerals or gatherings. Aid organizations may reroute convoys. People displaced by one conflict can become displaced again after a strike. The reported flight of residents from Jari Number 2 and fear of retrieving bodies in Kiremu illustrate how the psychological and humanitarian effects extend beyond the number of deaths on a particular day.
Ethiopia’s drone war has developed from a military innovation into a recurring feature of several conflicts. The latest reports show that civilian protection cannot be evaluated through claims of precision alone. It requires independently investigated incidents, transparent targeting rules and reliable records that distinguish verified casualties from allegations, and drone strikes from other air attacks.
Methodology and editorial note
This article is an expanded, number-rich editorial adaptation of the user-provided research file ‘Pasted text(9).txt’. It has not independently re-verified each underlying report. All counts are attributed to the reporting summarized in that file; conflicting numbers are retained as ranges. The article avoids adding casualty estimates across potentially overlapping reports and does not attribute unidentified attacks to a specific actor without support.
Selected sources identified in the supplied research
• Human Rights Watch — investigations of Dedebit and other airstrikes in Ethiopia
• BBC Amharic and BBC Tigrinya — reporting on Jari Number 2 and Kiremu (October–November 2026)
• AFP and Anadolu Agency — reporting on Rama and Jari Number 2 (October 2026)
• ACLED — conflict event reporting for September–October 2026
• Action on Armed Violence — civilian harm from explosive weapons in Ethiopia
• Insecurity Insight; Physicians for Human Rights / OJAH — attacks on health care and health-worker survey