Amharic Language Largely Invisible to AI Moderation Systems, Study Finds

Amharic Language Largely Invisible to AI Moderation Systems, Study Finds

Despite being the official working language of Ethiopia and spoken by over 40 million people, Amharic remains largely invisible to artificial intelligence systems that moderate content across major social media platforms, according to research compiled by Global Voices and academic studies on large language models.

A 2025 study, “The State of Large Language Models for African Languages,” comparing major AI systems found that only four African languages, Amharic, Swahili, Afrikaans, and Malagasy, are handled with any degree of consistency. That leaves more than 98 percent of Africa’s 2,000 languages essentially invisible to moderation systems that decide what content stays up and what gets removed.

Bereket Tsegay, an Amharic speaker hired to moderate content on TikTok at the company’s Kenya hub, told the Christian Science Monitor that he spent his days watching videos in languages he did not understand, including Luo, Dholuo, Kikuyu, and Dinka. When nothing in the visuals looked obviously wrong and no user had reported the video, he usually left it up. He has since left the job.

In Ethiopia, false claims circulating on Facebook alleged that Ethiopian troops had seized Eritrea’s Red Sea port, spreading widely before being debunked by fact-checkers. Fact-checkers interviewed by Poynter described spending hours manually tracking Amharic-language Facebook posts during periods of political tension in Ethiopia, doing work that should have been caught by the platform’s systems.

When a moderation system cannot process a language, it is less likely to flag content for human review. Instead, it relies on indirect signals such as user reports, visual cues, or audio patterns from languages it does recognize. This creates two problems: false positives, where content is wrongly removed, and false negatives, where harmful content in languages the system cannot parse stays up.

Mercy Mutemi, executive director of the Oversight Lab, a Kenyan legal advocacy group focused on technology, stated: “We are talking about an algorithm trained predominantly in English, being trusted to take down harmful content, while a huge percentage of TikTok users in Kenya are using TikTok in their mother tongue.”

The EU AI Act, which entered into force in August 2024, requires that AI systems be non-discriminatory and that training data be representative of the populations the system will affect. The Digital Services Act requires platforms to explain content moderation decisions to affected users. If a system cannot identify the language a post is written in, it cannot produce a meaningful explanation for why that post was removed.

Ethiopian researchers and academic teams at universities in Addis Ababa are working to build training data for languages that have almost none. The African Union’s Continental AI Strategy, approved in July 2024, commits to a people-centric approach and names data sovereignty as a priority, but strategy documents alone do not close the gap.

SourceEBR