On March 14, 2026, Adonay Berhane, Ethiopia’s most-followed TikTok creator with 5.4 million followers, posted what he described as potentially his “last video” on the platform. His offense? Sitting down and “speaking calmly” to his audience about life, growth, and healing. Adonay’s account had been hit with an unprecedented wave of community guideline violations, resulting in mass content removals. In a direct appeal to TikTok leadership, he described what many Ethiopian creators now recognize as a systemic problem: coordinated mass reporting campaigns designed to trigger automated content moderation systems and silence legitimate voices. Adonay is not alone. Across Ethiopia’s TikTok community, creators with audiences ranging from thousands to millions report sudden account suspensions, shadowbans, and content removals that they attribute to mass-reporting attacks. The pattern is consistent: accounts with no history of violations suddenly face cascading enforcement actions, often coinciding with competitive rivalries or content that challenges certain narratives.

Perhaps most disturbing are the persistent claims of an alleged “Reporting-as-a-Service” market, an underground economy where individuals purport to offer account takedowns, mass comments, and coordinated reporting for hire. Multiple creators openly discuss the belief that people can purchase everything: reports, comments, engagement, and even account suspensions. These allegations require verification, but their consistency suggests a phenomenon worthy of urgent investigation…The core problem lies in how platforms scale content moderation. With billions of posts daily, platforms rely heavily on artificial intelligence to detect violations and automate enforcement. But AI content moderation systems have well-documented limitations that are particularly acute in the Ethiopian context. Primarily, language barriers create fundamental challenges. Most AI models work best with high-resource languages like English and Mandarin, where developers have access to vast amounts of training data. Amharic, Tigrigna, and Afaan Oromo, Ethiopia’s primary languages, have limited training data due to their status as low-resource languages. Accuracy in detecting context, tone, and intent suffers dramatically as a result. Cultural context compounds these technical limitations. AI struggles with context-dependent content, where a word or phrase that’s benign in one cultural setting may be inflammatory in another. Without local expertise embedded in moderation workflows, platforms cannot distinguish between legitimate expression and genuine violations. A reference that carries deep political meaning in Ethiopian discourse might appear innocuous to an algorithm trained on Western datasets. These vulnerabilities make AI systems particularly susceptible to coordinated manipulation. Mass reporting campaigns exploit the weakness by mimicking the signals platforms use to identify violations: high report volume, rapid escalation, pattern consistency…

The weaponization of mass reporting represents a fundamental challenge to how we govern digital public spheres in politically complex environments. When platforms allow their moderation systems to become tools of censorship and economic warfare, they undermine the very communities they claim to serve. When AI moderation operates without local expertise, it becomes a vector for manipulation rather than a safeguard against harm. Ethiopia’s digital creators deserve platforms that understand their context, protect their livelihoods, and resist manipulation by bad actors. The technology exists. The expertise exists. What’s missing is the institutional will to deploy both to advance fairness and transparency in one of the world’s most challenging media environments. The current trajectory is unsustainable. The stakes are too high to accept systems that actors can purchase, manipulate, and deploy as instruments of silencing.