Facebook Algorithms Fail to Detect Hate Speech in African Languages, Fueling Violence: Analyst

© AP Photo / Matt RourkeThe Facebook logo is displayed on an iPad in Philadelphia, May 16, 2012
The Facebook logo is displayed on an iPad in Philadelphia, May 16, 2012 - Sputnik Africa, 1920, 02.05.2026
Subscribe
Facebook Algorithms Fail to Detect Hate Speech in African Languages, Fueling Violence: Analyst

Meta’s* content moderation systems are "linguistically blind" to harmful speech in many African languages, allowing calls for violence to go undetected while platforms respond quickly to similar content in Europe, cybersecurity analyst Anna Andersen told Sputnik Africa.

Andersen, an expert with the Global Fact-Checking Network, pointed to three root causes:

underrepresentation of African languages in training data;

complexity of local linguistic structures;

lack of cultural intelligence in centralized moderation models.

Kenya’s High Court ruling allowing Meta to be sued over ethnic violence amplification marks a potential turning point, Andersen noted.

Without sustained legal and political pressure, platforms have little incentive to shift from profit‑driven systems to context‑sensitive moderation, she warned.

*Meta is banned in Russia over extremist activities

Subscribe to @sputnik_africa

Sputnik Africa | X
Newsfeed
0