TikTok removed 750,185 videos in Kenya between April and June 2026 for violating its Community Guidelines, including content involving sexualised behaviour, dangerous activities, child abuse, harassment, scams and misinformation.
The removals have raised questions about whether moderation systems can identify harmful content without restricting legitimate information.
The platform also removed more than 46,000 accounts suspected of belonging to children below the age of 13, while 80,640 LIVE rooms were interrupted for violating its guidelines.
Some Kenyan creators, however, say automated moderation fails to distinguish harmful material from content intended to educate or inform.
Sheign Derrick, popularly known as Teacher Sheyii, says sexually explicit material, nudity, fraudulent LIVE broadcasts and inappropriate challenges involving children are particularly harmful to younger users.
“The erotic kind of content, the nudity kind of content are not safe, especially for children,” he told the
Sheign Derrick, popularly known as Teacher Sheyii, is a Kenyan TikTok content creator and political commentator known for his social and political activism on digital platforms.
Photo credit: Pool
He says such videos attract large audiences, creating financial incentives for creators.
“It gives somebody a lot of views. So it’s lucrative to do those kinds of content,” he said.
Derrick is also concerned about fraudulent broadcasts promising unrealistic returns. He recalls a TikTok user offering followers Sh3,000 in exchange for Sh1,000, warning that such schemes ex-ploit people seeking financial opportunities.
TikTok’s global figures show the scale of the challenge. Regulated goods, services and commercial activities accounted for 68.4 million removals, representing 34 per cent of all videos taken down. Sensitive and mature themes accounted for another 39.3 million, while mental and behavioural health violations accounted for 32.4 million.
Mental health violations included dangerous challenges, suicide and self-harm content, and dis-ordered eating.
Another 27.5 million videos were removed for safety and civility violations, including harassment, bullying and hate speech. Youth safety accounted for 23.8 million removals, while misinformation formed part of 6.8 million integrity and authenticity violations. Privacy and security violations accounted for another 3.5 million removals.
But while creators acknowledge the need to protect users, some say moderation can also penalise legitimate educational and public-interest content.
Cynthia Njue, a reproductive health content creator popularly known as Housegirlfriend, says she was subjected to cyberbullying after someone recorded her in private without consent and circulated the footage online.
Cynthia Njue, popularly known as Housegirlfriend, creates reproductive health awareness content on social media.
Photo credit: Pool
She reported the video, saying it had been recorded without her permission and was being used to facilitate harassment, but TikTok informed her it did not violate its guidelines.
In June 2026, she found that five videos providing information on safe sex and reproductive health services had been restricted.
“I appealed, asking them to consider the videos in the context of my work as a reproductive health creator,” she said.
For almost a week, she contacted TikTok’s customer service but says she received only automated responses and was never connected to a human moderator.
The appeal failed, forcing her to recreate videos and find other ways to communicate sensitive information. She says words such as ‘sex’ can trigger moderation even in educational content.
“We have to find some discreet ways to go around the violations because sometimes they can’t get the context,” she said.
Even when an appeal succeeds, she says, the initial removal affects a video’s distribution because it loses the momentum it would have gained immediately after publication.
Derrick, also a political commentator, says moderation often overlooks African cultural and social contexts.
He recalls a video showing a political leader addressing supporters while standing on top of a vehicle during a rally being flagged for promoting harmful activities.
“I’m a reporter. I go for a rally, I record some leader speaking on top of a vehicle. TikTok pre-sumes that the person who is on top of a vehicle is actually violating a community guideline,” he said.
He also experienced the removal of a video reporting the death of Patrick Marshall in September, resulting in a warning against his account. The warning was later withdrawn after the platform established that the information was accurate, but the video remained restricted. “What if I am the first person to break that particular news? Where is the problem?” he asked.
He says slow appeals can undermine the reach of breaking news. For both creators, the challenge is enforcing safety guidelines without suppressing legitimate journalism, education and public-interest discussions.
Njue wants TikTok to invest in human moderators who understand local languages, including Kiswahili, Sheng and indigenous languages, to interpret content more accurately.
“If they’re going to use AI, let them train the AI to look at context. If they’re going to use hu-mans, the better. Let them look at the context in which a video has been created,” she said.
Odanga Madung, Managing Director at Odipo Dev, says the shift from traditional print and broadcast media to mobile platforms has made information consumption more personalised, faster and increasingly driven by peer-to-peer sharing.
“It has changed the nature of the infrastructure by which people consume information. It is now heavily individualised, mobile, fast and peer-to-peer,” he said.
He says young people favour TikTok’s video content because it reflects their interests. Algorithms shape what users see by promoting some material and limiting other content.
“Algorithms are values embedded in code,” Mr Madung said. “These systems influence both what content is allowed to circulate on a platform and what is recommended to individual users.”
He warns that platforms’ growing role as information gatekeepers affects political discourse, online scams and health decisions.
TikTok says its enforcement efforts combine automated moderation tools with the expertise of thousands of trust and safety professionals worldwide.
During the second quarter, automated systems detected 97.4 per cent of videos removed globally, while approximately 6.3 million videos were reinstated following successful appeals.
In Kenya, nearly 11,000 videos were removed under policies covering edited media and artificial intelligence-generated content. TikTok says it has introduced creator labelling tools, automated detection and C2PA Content Credentials technology to help users identify content created or significantly altered using artificial intelligence.
The figures highlight the challenge for platforms: protecting users while ensuring automated enforcement does not silence education, journalism and public interest discussion.