As military tensions escalate between Ethiopia and Eritrea, AI-generated images and videos are inflaming animosity online.
Since October 2023, Prime Minister Abiy Ahmed has repeatedly insisted that landlocked Ethiopia must have direct sea access, with Eritrea’s port of Assab frequently cited as a potential target.
Tensions have been rising in recent weeks, with reports of troops being moved towards their shared border.
That has been matched by increasingly violent rhetoric online, including a flood of images generated by artificial intelligence that seek to portray their side’s military dominance and humiliate rival leaders, while presenting war as swift and cost-free.
They are drawing thousands of interactions within hours of being shared, with comment sections full of threats and more violent imagery.
– ‘National narrative’ –
Eliyas Kebede Zemedkun, an Ethiopian with more than 80,000 Facebook followers, has spent months creating AI-generated images such as tanks entering Assab port to cheering crowds, or Ethiopia’s army chief chasing Eritrean President Isaias Afwerki.
He told AFP he wants to promote “Ethiopia’s national narrative” and challenge those that try to “downgrade the national army”, yet he also admitted that AI “obscures reality” and “normalises aggression”.
Eliyas said he uses free platforms like Gemini and ChatGPT to create the content, along with editing software Clideo.
Each post has triggered waves of retaliatory content from pro-Eritrea accounts.
One frequent user, under the name Mimta Grlis, has posted depictions of Abiy being arrested by Eritrean forces or an Ethiopian ship sinking in the Red Sea.
Kjetil Tronvoll, an expert on the region at Oslo New University College, said similar online clashes helped drive aggression during the civil war in Ethiopia’s Tigray region in 2020-2022, which killed an estimated 600,000 people, but this has been turbo-charged by AI.
“When people believe fabricated visuals are real, it fuels genuine anger, fear, and animosity,” he said.
Digital literacy is still limited in Ethiopia, which ranked 112 out of 149 countries on the latest World Economic Forum’s Digital Skills Index.
“The videos are not close to reality in most cases. But even when they’re not realistic, the emotional reaction is very strong due to their provocative nature and limited media literacy (of the viewers),” said AI expert Amanuel Meseret.
A review of the comments posted beneath the AI-generated images suggests many social media users believe they are genuine.
Decades of conflict and trauma also mean even obviously fake images can trigger audiences or play into their existing worldview.
– Divided opinion –
Alemshet Moges, an ex-military officer turned journalist, said a clip of a drone attack on the Eritrean city of Asmara reflected how he imagined a conflict could play out.
“I don’t want war to start, but if it does, I believe Ethiopian drones will be the game-changer,” he said.
Others are simply horrified.
“They are very harmful in our context,” said Hawi Abebe, an accountant in Addis Ababa when shown some of the content.
Workineh Diribsa, a journalism lecturer at Ethiopia’s Jimma University, said tech companies should be pressed to remove harmful images and that media literacy programmes were vital to help citizens “identify, question, and resist manipulative content”.
“These videos dramatise war as a quick and effortless victory… constructing a false reality that risks steering opinion and political discourse toward confrontation rather than resolution,” he said.
AFP