Online hate speech falls under the same law as offline speech in Kenya, but it's harder to investigate. The Inspector General of Police told the Senate in 2026 that foreign-hosted accounts, encrypted platforms, viral amplification and coordinated misinformation all complicate enforcement. NCIC's chairperson has said that a clip may be generated by artificial intelligence, so the commission needs a formal complaint and proof that the content is real.
Key points:
- Section 13 already covers publishing and distributing material.
- A proposed amendment would name electronic and social media.
- Foreign-hosted and anonymous accounts make enforcement difficult.
- Forwarding can spread harm, and it may also expose you legally.
The law applies online
Section 13 of the National Cohesion and Integration Act covers a person who publishes or distributes written material or shows a recording, and Article 33 draws no line between online and offline speech. The 2022 Amendment Bill proposed clearer words: "publishes, posts or distributes material in print, electronic or social media". The Computer Misuse and Cybercrimes Act adds its own offences.
Why is it hard to enforce?
The Inspector General named several reasons: accounts hosted outside Kenya, encrypted communication platforms, politicisation of enforcement, viral amplification and coordinated misinformation. An anonymous account may be run from another country, and a message in a private group leaves little that an investigator can see.
What about AI-generated clips?
Fake audio and video are now easy to make. NCIC's chairperson raised this directly: "we don't know whether the clip that was sent was AI-generated". A monitor should treat any striking clip as unverified until it's confirmed, by finding the original, the place and the people who were there.
Does forwarding count?
Section 13 covers those who distribute material, as well as those who create it. So forwarding a hateful message in a group may carry risk. Even if it doesn't, it spreads the harm. If you want to report it, send it to the authority, and don't forward it to friends.
What can platforms do?
Platforms have rules against hate speech, and ways of reporting it. Use them, and keep the link and a screenshot, with the date. Reporting in local languages can be less effective, because automatic tools work best in English. That's why local monitors matter.
How should a monitor track online hate speech?
Work from public sources, and use clear rules:
- Decide what you're looking for, and write it down.
- Include Kiswahili, Sheng and local languages.
- Record the link, the date, the words and the reach.
- Have a person confirm each item.
- Report patterns, not just incidents.
Be careful with lists of slurs. Publishing them teaches them, and a keyword list alone will produce many false alarms, because people often use the same words to condemn hate speech.
Should you monitor private groups?
Private groups raise serious privacy and legal questions. A public body needs a clear legal basis for monitoring them, and people's consent where it applies. Monitoring should focus on public speech, and on complaints people choose to make.
How should a monitor protect people?
Never build lists of people by community. Report what was said, and leave out who belongs where. A system that counts things against ethnic groups can become the kind of tool it's meant to prevent. Keep personal details safe, and don't publish names of private individuals.
How do rumours spread before violence?
Rumours spread fastest when people are afraid. A false claim about an attack can move in hours from a post to a crowd. A quick, trusted correction on the same channels is the best response. Conflict early warning sets out how to respond.
What can an ordinary user do?
Pause before you share. Ask who made this, and whether it's been reported elsewhere. Don't add insults in reply. Report it. If you're in a group, say calmly that the claim is unverified.
What should a county do?
Tell residents where to find correct information, and be quick to correct falsehoods about county matters. Work with local radio, since many people check rumours there.
What should happen before 2027?
Platforms, the police, NCIC and civil society each hold part of the answer. NCIC's chief executive has spoken of upgrading digital platforms, and of concern about new forms of harmful communication. Clear rules and fast reporting channels will matter more than slogans.
What has changed in the law in 2026?
Two rulings matter. The Court of Appeal struck down the false publication offences in March, and the High Court struck down cyber harassment and website blocking provisions in July. An appeal is before the Supreme Court on the first.
Section 13 of the National Cohesion and Integration Act still applies to hateful content, wherever it's published.
Who is paying for this content?
Some of it is paid for. A Mozilla investigation found influencers paid Sh1,000 to Sh1,500 to join three Twitter campaigns a day, and Twitter suspended just over 100 Kenyan accounts. Global Voices interviewed people hired as online influencers in 2022.
So don't assume that a trending hashtag shows what people really think.
What is the scale?
Mozilla's 2022 report found 130 TikTok videos that pushed disinformation and stoked ethnic discriminatory narratives, and they drew millions of views. That's one sample on one platform, and it doesn't show the whole picture.
What should a parent or teacher do?
Talk to young people about what they're seeing. Ask who made it, and why. Show them how to report a post. NCIC's Amani Clubs in schools teach peace, diversity and responsible citizenship, and they're a natural place for this talk.
Read next
- Conflict early warning system for Kenya: the TI pack page for this work
- Hate speech in Kenya: all the pages in this section
- Dangerous speech
- Computer Misuse and Cybercrimes Act
- Monitoring cohesion without profiling people
- Takwimu Intelligence (TI) home page
- National cohesion in Kenya: the guide to this library
- The TI platform and how we check our numbers
Where TI fits
Takwimu Intelligence (TI) tracks public media in several languages and keeps every report unverified until a person confirms it, with no lists of communities.
To see it on your own work, book a walk-through.