Kenya Freedom Index - Deep Dive
Deaths,
Injuries,
Arrests
& Weapons
A data-driven examination of the human cost of protest in Kenya, 2025–2026
How Many Kenyans Never Made It Back Home After Protests?
In 2025, the protest hot months of June and July were the deadliest of the year. Those who died were protesting against police brutality and impunity alongside governance failures. In 2026 however, the recorded deaths sit at 7.
Innocent Bystanders are also at risk
Not everyone who dies during a protest chose to be there. Bystanders, residents near protest routes, shopkeepers, children in their homes, and passers-by have all been recorded among protest-related deaths in Kenya.
Boniface Mwangi Kariuki, 22, was a mask vendor working in Nairobi's central business district. On 17 June 2025, a police officer shot him in the head at close range and he was then declared brain-dead at the Kenyatta National Hospital. He died on 30 June. Bridgit Njoki was twelve years old when she was killed on 7 July 2025 during the Saba Saba protests. She was watching television in her living room when a single bullet pierced the roof of her home, punctured the ceiling and struck her in the head. She was pronounced dead within hours. Fred Wamale Wanyonyi, a security guard on duty at a mall in central Nairobi, was killed on 25 June 2025 while simply doing his job.
These deaths deserve their own count and their own reckoning. They show that the cost of how Kenya polices the protests is borne not only by those who protest, but by the communities in which protests occur. Every bystander's death is a measure of how far state force has overreached its mandate.
What Were the Total Number of Injuries? (2025-2026)
June 25, 2025 was the most physically harmful single day of protest in this dataset. The injuries recorded on that day alone exceeded every other protest date. July 7, 2025 added further injuries on top of an already significant toll, making the June and July 2025 period the most damaging time for protesters. 2026 so far shows a much lower injury count from the recorded protest hot months.
What Were the Nature of Injuries in 2025?
Across the June 25 and July 7, 2025 protest hot months, 58 injuries were recorded as gunshot wounds, by far the largest category. These should not be argued as accidental discharges or merely collateral consequences of trying to disperse a crowd. They are uses of lethal force against civilians who were, in most cases, exercising a constitutional right. Teargas canisters caused 8 recorded injuries. This is significant because these occur when the metal canister itself hits a person at high velocity, not only when gas disperses. A further 8 injuries were recorded as physical assault, 9 as other body injuries and 27 recorded without a specified type.
When The Helpers Get Attacked- Medics, Journalists, Civil Society
Those who go to a protest to help, document, or observe should not become targets. However, in Kenya they repeatedly do. Medics treating the wounded have been teargassed and shot at. Journalists covering protests have been arrested, beaten, and had their equipment confiscated. Civil society workers monitoring human rights have been detained and intimidated.
Targeting helpers is not incidental. It is strategic. When medics cannot work, the injured suffer longer. When journalists cannot report, accountability is harder to establish. When civil society cannot monitor, abuses go unrecorded. Each attack on a helper is an attack on the infrastructure of accountability itself.
What Were the Total Number of Arrests? (2025-2026)
In 2025, the peak arrest day was July 7 during the Saba Saba protests. In 2026, the two dominant recorded protest days produced approximately 1,000 arrests. June 25th appears in the data as a consistent arrest flashpoint, with hundreds detained on that date in both 2025 and 2026.
How was this data collected?
We defined a protest as a collective public action expressing disapproval of, objection to, or demand for action on a policy, institution, authority, condition, or public issue. Following this definition, we included as a protest any form of march, go-slow, picket, demonstration, sit-in, riot, strike, work stoppage, or confrontation involving the police. This is consistent with the right guaranteed under Article 37 of the Constitution: "Every person has the right, peaceably and unarmed, to assemble, to demonstrate, to picket, and to present petitions to public authorities."
This dataset is the product of a three-stage approach combining automated and manual research. Stage 1 (automated search) used code to run targeted searches against Google's RSS/News index, returning articles matching our keyword specifications; the resulting set was then human-reviewed and deduplicated. Stage 2 (manual search) filled in protests that don't get reported as news or aren't indexed in Google's RSS feed, using a search string built around terms including "protest," "demonstration," "maandamano," "riot," "unrest," "picket," "go slow," "strike," "boycott," "teargas," "clashes," "barricade," and "took to the streets." Stage 3 (YouTube search) used a parallel search string to catch protests that don't make it into print or standard news coverage.
Data on deaths, injuries, and arrests is sourced from the Independent Policing Oversight Authority (IPOA), the Kenya National Commission on Human Rights (KNCHR), Amnesty International Kenya, and the Ministry of Interior & National Administration. These bodies do not always report on the same incidents, and when they do, their figures sometimes differ: for example, KNCHR and IPOA have each published different counts of arrests and injuries on the same date. Where more than one source reports on what is clearly the same protest period, we use the highest figure reported for that period rather than adding the figures together, since summing competing counts of the same events would overstate the true total. Figures for non-overlapping periods are added together as normal. Deaths, injuries, and arrests counters on this page reflect specific high-impact "hot month" dates called out by these sources (not a comprehensive year-round tally) so they are presented as "at least" figures.
Where available, protest locations were identified through direct references in the news articles. It was not possible to geotag or report all protests with precise geographic coordinates. In such cases, the nearest identifiable location, such as a constituency, sub-county, or another recognisable administrative area, was used. Coordinates were then assigned based on these approximated locations for mapping purposes.
For each documented protest, we aimed to identify whether there was any official response from authorities or public institutions. Some protests did not elicit any official response at the time the stories were published. In such cases, the dataset reflects the absence of available responses rather than assuming that none occurred.
Certain protests took place over multiple days, making it difficult to capture exact start and end dates with precision. For these cases, reported timelines were used as faithfully as possible, although discrepancies in reporting styles across media outlets remain a known limitation.
Want to record a protest for us to consider? Click here and share the date, location, organiser, purpose, and at least two sources of information.
Got suggestions on how we can improve the Freedom Index? Email us team@odipodev.com.
Except where otherwise noted, content in this website is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International licence: https://creativecommons.org/licenses/by-nc-nd/4.0/legalcode. Where material is attributed to a copyright owner other than Amnesty International or Odipo Dev, that material is not subject to the Creative Commons licence.