Dangerous speech is speech that makes people more likely to accept or take part in violence against another group. The Dangerous Speech Project says that leaders use it to convince their supporters to perceive other groups as monstrous threats, so that mass violence seems not only acceptable, but necessary. It differs from hate speech, which is a legal category in Kenya. Dangerous speech is a way of judging the risk of harm.
Key points:
- It focuses on the likely effect of speech, more than on how offensive it sounds.
- The most dangerous speech often sounds like self-defence.
- Who speaks, and to whom, matters as much as the words.
- It can be countered.
What is dangerous speech?
The project describes it as a special kind of rhetoric that leads people to think that mass violence is acceptable and even necessary. It works by fear. A group is shown as a threat to "us", so harming them seems like protection.
How does it differ from hate speech?
Hate speech is often defined by content, such as insults or contempt. Under Kenyan law, it's a crime with set elements. Dangerous speech asks a different question: will this make violence more likely? Effect matters more here.
So some speech can be polite and still dangerous. A calm statement that "they are planning to attack us first" may do more harm than a crude insult. Equally, an insult may be illegal without making violence likely.
How can you judge the risk?
Five plain questions help:
- Is the speaker influential with this audience?
- Is the audience afraid, or does it have grievances?
- Does the message present another group as a threat?
- Is there a history of violence between these groups?
- Is it being spread widely, including in local languages?
The more of these apply, the higher the risk.
What are the warning signs?
Watch for messages that describe people as insects, animals or a disease, that claim the other group is about to attack, that say that "we" must defend ourselves, and that say the group doesn't belong here. Kenya's Sessional Paper warns of calls to exclude "intruders". Any of these can be a warning sign, whether or not it's illegal.
Why does this matter in Kenya?
Past election violence followed months of rising tension. Monitors who look only for illegal words may miss the build-up. A focus on dangerous speech helps peace actors respond before anything breaks the law.
How should a monitor use this?
Don't try to rank every post. Focus on the speakers and places with the highest risk. Record the exact words, the language, the date and the context. Label your view as a judgement, and keep it separate from the evidence. Never publish a list of speech that repeats the harm.
How can it be countered?
The project works on counterspeech, and says that dangerous speech can be countered. Good responses come from people the audience already trusts: elders, religious leaders, teachers and local radio hosts. Calling out the claim, offering true information and showing shared interests can all help. Responding in a way that spreads the original message can make things worse.
What should a peace committee do?
When a rumour starts, get facts quickly. Bring leaders of both groups together, and let them speak publicly. Tell people where to report. Use radio. Conflict early warning shows how to move from noticing to acting.
What should media do?
Avoid repeating fear claims without checking them. Ask "who says so?" and "what is the evidence?" Don't let a live caller make an unchecked claim about another community. Give the correction the same prominence.
What about the law?
Dangerous speech isn't an offence in itself. Where it amounts to incitement, hate speech or ethnic contempt, the law applies. Where it doesn't, the answer is more and better speech, and that respects freedom of expression.
What are the limits of the idea?
Judging risk is hard, and people disagree. Two honest monitors may read a statement differently. Use a clear checklist, have two people review, and record why you reached your view. Treat it as a guide for prevention, never as proof of guilt.
What does this look like in Kenya today?
NCIC monitoring has detected a resurgence of coded ethnic language and incitement online and offline. NCIC warned earlier of ethnic balkanisation and zoning by politicians. In 2026, a civil society speaker warned against politicians being barred from an area because it doesn't belong to them.
Language that says a group doesn't belong is an old warning sign. Listen for it.
How do leaders' words travel?
A remark at a rally reaches radio by evening and phones within minutes. The Media Council ordered a seven-second delay on live broadcasts in 2026, because unchecked rally coverage had carried inflammatory statements. Online, there's no such delay.
Those with the greatest reach carry the greatest duty. That's why a speaker's standing matters so much in judging risk.
What should faith leaders do?
Name the danger without repeating the words. NCIC has told church leaders that they have the authority to stop politicians from using worship centres to spread divisive messages. Preach about neighbours before a rally comes to town. Afterwards, go to the people who felt targeted.
How should you keep a record?
Note the speaker, the place, the date, the audience, and the exact words in the original language. Add what happened next. Records like this show patterns over time, and they can later support a complaint. Hate speech monitoring gives the method.
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
- What is hate speech in Kenya?
- Online hate speech
- What is conflict early warning?
- 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) keeps the evidence, the reading of it, and who made that reading apart, so a judgement never passes for a finding.
To see it on your own work, book a walk-through.