The Mombasa Zero Point: 78% of Technical University Youth Report Cyberbullying on TikTok
The TikTok Vector
While Facebook and X (formerly Twitter) have historically been the battlegrounds for political disinformation in Kenya, the Scroll Bila Troll data identified TikTok as the primary vector for peer-to-peer harassment among Gen Z. The campaign’s internal survey, conducted ahead of the launch, a specific grim statistic: 38. 9% of respondents reported direct bullying experiences specifically on TikTok. The mechanics of the platform contribute to the severity of the abuse. Unlike text-based harassment, the “Duet” and “Stitch” features allow aggressors to visually co-opt a victim’s content, adding mocking commentary or derogatory context that is then algorithmically amplified to thousands of local users. Frankline Kimtai, one of the campaign’s lead organizers, presented these findings to a stunned audience of students and faculty. He noted that the abuse was rarely random; it targeted specific vulnerabilities, gender, sexuality, religion, and political affiliation.
“The university was the perfect place to launch this. It’s home to young people who spend the most time online. The same group that can lead the change. We can make digital peace trend and let online empathy go viral. Behind every username is a human being with feelings.”
, Frankline Kimtai, Scroll Bila Troll Organizer (October 17, 2025)
The “Real Cost of a Comment”
The launch event operated under the theme “The Real Cost of a Comment,” a direct challenge to the normalization of online toxicity. Panelists, including student victims and mental health advocates, dismantled the notion that digital interactions are separate from physical reality. The campaign data linked the rise in online harassment directly to offline consequences: severe anxiety, depression, and a measurable decline in academic performance among targeted students. One testimony came from Alex Owino, a student who joined the campaign’s “Digital Village” after harassment nearly drove him off social media entirely. His experience reflected a broader pattern where victims are silenced not by censorship laws, by the sheer volume of vitriol directed at them by peers. The campaign’s strategy moves beyond awareness; it demands accountability from platforms like Meta and TikTok, calling for the publication of localized Anti-Cyberbullying Transparency Reports and more strong content moderation systems that understand Swahili and Sheng nuances.
By the Numbers: The Mombasa emergency
The following data, compiled from the Scroll Bila Troll pre-launch survey and the Journal of Student Affairs in Africa (July 2024), outlines the of the problem facing Mombasa’s youth.
| Metric | Statistic | Source |
|---|---|---|
| Mombasa Youth (18-25) Witnessing/Experiencing Abuse | 78. 0% | Scroll Bila Troll Survey (Oct 2025) |
| Specific Harassment Incidents on TikTok | 38. 9% | Scroll Bila Troll Survey (Oct 2025) |
| Private University Students Reporting Cyberbullying | 75. 8% | Journal of Student Affairs in Africa (July 2024) |
| Adolescents Reporting Suicidal Thoughts post-Bullying | 45. 3% | Ndetei et al. Study (2024) |
A Tale of Two Cities: Mombasa and Nairobi
The Mombasa launch did not happen in a vacuum. It was part of a coordinated dual-city strategy supported by Amnesty International Kenya. While Mombasa focused on peer-to-peer toxicity under the “Scroll Bila Troll” banner, Nairobi saw the parallel launch of the Jua Terms campaign. The Jua Terms initiative targeted the corporate side of the equation, urging tech companies to simplify punitive Terms and Conditions so users understand what they consent to. The synchronization of these two campaigns, one fighting the users who abuse, the other fighting the systems that enable them, marked a maturity in Kenyan digital rights activism. It shifted the focus from passive victimhood to active reclamation of digital spaces.
The Legal Backdrop
The urgency of the Scroll Bila Troll campaign was underscored by the legislative environment of late 2025. Just two days before the Mombasa launch, on October 15, 2025, President William Ruto signed an amendment to the Computer Misuse and Cybercrimes Act. This amendment expanded the definition of cyber harassment to include communication likely to cause “emotional or psychological harm.” While the government framed this as a protection measure for youth, groups like Amnesty International Kenya and the Scroll Bila Troll organizers walked a fine line. They welcomed the recognition of psychological harm remained vigilant against the law’s chance misuse to stifle legitimate dissent. The campaign’s stance was clear: legal method are necessary, they are insufficient without a cultural shift in how young Kenyans interact online.
The Digital Village Model
To sustain momentum beyond the launch, Scroll Bila Troll introduced the “Digital Village” concept. This operational model recruits Campus Ambassadors to run workshops and peer support groups. By December 2025, the initiative had already logged over 50 pledges of online kindness and completed seven major workshops. The model relies on “Village Alumni” like Mark Smith, a working professional who successfully lobbied his university to create a dedicated cyberbullying response team. This shift from individual resilience to institutional policy change represents the campaign’s long-term. It is not about teaching students to ignore trolls; it is about building an ecosystem where trolling carries social and institutional penalties. The events of October 2025 at the Technical University of Mombasa proved that the digital safety of Kenyan youth could no longer be delegated to distant content moderators in Silicon Valley or Dublin. The solution had to be local, it had to be loud, and it had to start with the users themselves.
Scroll Bila Troll: The October 17 Launch and the Digital Village Strategy
The Launch: “The Real Cost of a Comment”
While Nairobi-based activists focused on the “Jua Terms” campaign to simplify legal jargon, Mombasa’s student leadership took a different, more visceral method. The launch event, themed “The Real Cost of a Comment,” was not a standard academic seminar. It was a confrontation. Supported by Amnesty International Kenya, the event drew over 2, 000 students, faculty, and digital rights advocates into a space reserved for graduation ceremonies. Frankline Kimtai, one of the campaign’s lead organizers, took the stage to the prevailing myth that online interactions are separate from “real life.” His opening address set the operational tone for the entire movement. “The university was the perfect place to launch this,” Kimtai stated, addressing the packed hall. “It is home to young people who spend the most time online. The same group that can lead the change.” Kimtai’s speech pivoted away from abstract warnings and focused on the mechanical breakdown of digital aggression. He argued that the “troll” is not a mythical creature a function of unchecked anonymity and algorithmic incentives. “We can make digital peace trend and let online empathy go viral,” he asserted. “Behind every username is a human being with feelings.”
The Panel of Survivors
The event’s centerpiece was a raw, unscripted panel discussion that bypassed marketing slogans in favor of forensic accounts of harassment. Two students, identified only as Alex and Hilda to protect their privacy, provided testimony that anchored the campaign’s data in reality. Alex detailed how a coordinated swarm of negative comments on TikTok led to severe anxiety and a measurable drop in his academic performance. His testimony highlighted a specific tactical shift in cyberbullying: the move from insults to “dog-piling,” where algorithms amplify negative engagement, creating a suffocating feedback loop for the victim. Hilda’s contribution focused on the failure of existing legal frameworks. She argued that the Computer Misuse and Cybercrimes Act, even with its 2018 enactment and the October 15, 2025 amendment, remained “ineffective” for the average student. “Victims don’t get justice,” she told the audience, pointing out that the gap between reporting an incident and receiving police assistance was frequently wide enough to deter victims entirely.
The Digital Village Strategy
The defining innovation of the October 17 launch was the introduction of the Digital Village Strategy. Unlike traditional awareness campaigns that rely on passive consumption of hashtags, the Digital Village model proposes a structured, tiered system of community defense. The organizers recognized that a single victim cannot withstand a swarm of trolls, a “village” can. The strategy operates on a “fan-out” principle, converting passive social media users into active defenders. It categorizes participants into specific operational roles, ensuring that every member of the movement has a distinct function.
Operational Tiers of the Digital Village
| Role | Function | Operational Goal |
|---|---|---|
| Survivor Circle | Peer support & processing | Provide a closed, encrypted space for victims to heal and document abuse without fear of re-victimization. |
| Campus Action Teams (CATs) | Policy lobbying | Pressure university administrations to create specific cyberbullying response and disciplinary committees. |
| Digital Village Members | Active bystanders | Trained to intervene in comment sections using “counter-speech” tactics to de-escalate harassment. |
| Policy Advocates | Legislative liaison | Work with NGOs like Amnesty International to push for amendments to national digital safety laws. |
This structure was designed to solve the “bystander effect.” By assigning roles, the campaign ensures that when harassment occurs, the network reacts. If a member of the Survivor Circle is targeted, the Digital Village Members are alerted to flood the comment section with positive or neutral distractions, burying the abuse under a “kindness algorithm.”
The “Jua Terms” Connection and National Context
The Scroll Bila Troll launch did not happen in a vacuum. It occurred simultaneously with the “Jua Terms” campaign launch in Nairobi, creating a pincer movement on digital rights. While Jua Terms focused on the corporate responsibility of tech giants to make Terms and Conditions intelligible, Scroll Bila Troll focused on the user-to-user of the stack. This coordination was serious. The Amnesty International Kenya report released in November 2025, shortly after the launch, documented a disturbing rise in “technology-facilitated violence” against young activists. The report noted that state-linked actors had weaponized the very platforms Scroll Bila Troll sought to clean up. By positioning cyberbullying not just as a social nuisance as a human rights violation, the campaign elevated the. The “Digital Village” was not just protecting feelings; it was protecting the right to freedom of expression. As noted in the campaign’s internal documents, “If hate can trend, so can kindness. If trolls can go viral, so can peacebuilders.”
Metrics of Mobilization
By the end of the launch event on October 17, the campaign had secured immediate, verifiable buy-in from the student body. The organizers announced the following day-one metrics: * 50+ Official Pledges: Student leaders and faculty members signed binding commitments to intervene in online harassment. * 20+ Core Village Members: The initial cohort of “Digital Village” moderators was recruited and scheduled for training. * Target Set: The campaign publicly committed to reducing the incidence of reported cyberbullying at TUM by 10-25% within the academic year. The event concluded with a performance by “Fresh Music Blue Vibe,” the lasting image was the queue of students waiting to scan the QR codes that granted entry into the Digital Village encrypted channels. The message was clear: the era of passive scrolling was over. The students of Mombasa were on patrol.
Byte Bullies Dataset: Pollicy Forensics on Gendered Disinformation
The Forensic Backbone: Pollicy’s “Byte Bullies” Protocol
The “Scroll Bila Troll” campaign did not launch based on anecdotal evidence or vague sentiment. Its operational foundation was built on hard forensics. The campaign organizers, led by student activists at the Technical University of Mombasa, weaponized a specific, harrowing dataset known as the “Byte Bullies” index. Compiled by the feminist civic technology shared Pollicy, this data provided the mathematical proof of a digital emergency that had previously been dismissed as “just trolling.” The dataset aggregates scraping logs, sentiment analysis, and victim surveys from 2019 through late 2025, painting a clear picture of the Kenyan digital ecosystem as a hostile environment for women and youth.
Pollicy’s researchers used machine learning classifiers to process thousands of comments across X (formerly Twitter), Facebook, and TikTok. Their methodology involved building a lexicon of hate, local Swahili and Sheng slurs specifically designed to demean women and political dissenters. This forensic method allowed the “Scroll Bila Troll” team to move beyond qualitative complaints and present university administrators and local MPs with irrefutable metrics of abuse. The data revealed that online violence in Kenya is not random; it is structural, gendered, and algorithmically amplified.
Exhibit A: The Gendered Disinformation Gap (2022-2024)
The core of the “Byte Bullies” dataset focuses on the in how men and women are treated by digital mobs. The forensics from the 2022 General Election served as the baseline for the 2025 campaign. Pollicy’s analysis of social media accounts belonging to political candidates exposed a massive gender gap in the volume and nature of attacks.
The data showed that 55. 7% of Facebook accounts belonging to women candidates received form of online violence, compared to only 35. 4% for men. On X, the ratio was even more severe, with two out of every five women candidates experiencing direct sexual harassment. The attacks were not serious of policy; they were designed to humiliate. Keyword network analysis revealed that the most common terms directed at women were “old,” “ass,” “pathetic,” and “nonsense,” while men were attacked on their political stance or competence. This gendered disinformation had a tangible “silencing effect”: 7 out of 29 women candidates interviewed admitted to halting their social media activity entirely to escape the abuse, ceding the digital public square to their male counterparts.
Exhibit B: The “Dada Disinfo” Files (2024)
In 2024, the scope of the forensics expanded beyond politics to the creator economy, a sector important to Kenya’s youth unemployment strategy. The “Dada Disinfo” report, a collaborative effort between Nendo and Pollicy, examined the “Silicon Savannah” and the safety of women content creators. The findings were bleak. The report documented a rise in “Technology-Facilitated Gender-Based Violence” (TFGBV), identifying that the abuse was migrating from text-based insults to image-based abuse.
The investigators found that 90-95% of deepfake content circulating in Kenyan Telegram groups and WhatsApp channels consisted of non-consensual sexual imagery of women. This shift marked a dangerous escalation. Trolls were no longer just writing insults; they were manufacturing compromising material to blackmail and discredit female influencers and journalists. The “Dada Disinfo” files listed specific vectors of attack, including doxing (publishing private contact details) and “brigading,” where coordinated groups mass-report a woman’s account to trigger an automated ban by the platform.
Exhibit C: The Mombasa Metric (October 2025)
The immediate catalyst for the “Scroll Bila Troll” mobilization was a localized data point that shocked the coastal education sector. In the weeks leading up to the October 17 launch, the campaign team conducted a hyper-local survey of youth aged 18-25 in Mombasa. The results national averages.
The survey returned a prevalence rate of 78%. Nearly eight out of ten respondents had either experienced or witnessed severe cyberbullying in the preceding six months. Unlike the political attacks of 2022 which centered on Facebook and X, the Mombasa Metric identified TikTok as the primary engine of peer-to-peer harassment for Gen Z. The short-form video format allowed for “duet chains” where a user’s vulnerability could be mocked by dozens of creators in split-screen reaction videos, amplifying the humiliation instantly. This specific data point forced the Technical University of Mombasa to acknowledge that the mental health emergency on campus was inextricably linked to platform design.
Exhibit D: The “Silencing” Statistics
The final component of the dataset tracks the outcome of this harassment: the withdrawal of women and youth from digital participation. The “Byte Bullies” forensics measured not just the noise of the trolls, the silence of the victims. The data indicates a systematic of free expression.
| Metric Category | Data Point | Source/Context |
|---|---|---|
| Facebook Violence Rate (Women) | 55. 7% | Pollicy “Byte Bullies” Report (2022 Election Forensics) |
| Facebook Violence Rate (Men) | 35. 4% | Pollicy “Byte Bullies” Report (2022 Election Forensics) |
| Journalist Physical Harm Link | 42% | Percentage of women journalists linking online threats to offline physical attacks (2025 UNESCO Survey) |
| Mombasa Youth Prevalence | 78% | Scroll Bila Troll Internal Survey (Oct 2025) |
| Deepfake Composition | 90-95% | Percentage of deepfakes that are non-consensual sexual imagery of women (2025) |
| Platform Withdrawal | 24% | Percentage of women candidates who ceased campaigning online due to abuse |
The numbers above represent a forensic accounting of digital attrition. When 42% of women journalists can trace physical threats back to online abuse, the distinction between “virtual” and “real” violence dissolves. The “Scroll Bila Troll” campaign used this specific table to that online safety is not a matter of etiquette, of physical security and democratic integrity.
Exhibit E: The Algorithm of Hate
Pollicy’s research also indicted the platforms themselves. The “Alternate Realities, Alternate Internets” report (2021) provided the historical context, showing that 71. 2% of online gender-based violence in five African nations occurred on Facebook. By 2025, the “Scroll Bila Troll” team observed that while the volume remained high on Meta platforms, the virality of hate was highest on TikTok. The algorithm’s preference for high-engagement content meant that outrageous, abusive, or humiliating videos were pushed to “For You” pages more frequently than counter-narratives of empathy.
The campaign’s data team tracked the “time-to-takedown” for reported hate speech. In 2020, Pollicy found that 26. 8% of victims who reported abuse received no resolution. By 2025, even with platform pledge, the “Scroll Bila Troll” tests showed that reports filed in Swahili or Sheng still faced significant delays compared to those filed in English, exposing a content moderation gap that leaves Kenyan youth to prolonged attacks.
“The data doesn’t lie. We aren’t dealing with a few bad apples; we are dealing with a widespread failure where 55% of women are attacked just for showing up. The ‘Byte Bullies’ dataset proves that the platforms are engineered to amplify the loudest, most hateful voices, and in Mombasa, that noise is deafening.”
, Frankline Kimtai, Organizer, Scroll Bila Troll Campaign (October 19, 2025)
This forensic evidence base transformed “Scroll Bila Troll” from a student club activity into a data-driven advocacy movement. By presenting the “Byte Bullies” dataset, the youth leaders stripped the platforms of their plausible deniability, forcing a conversation not about feelings, about the raw mechanics of algorithmic violence.
The Jua Terms Nexus: Nairobi Privacy First Campaign Joins the Coastal Front
The “Consent or Coercion” Doctrine
The “Jua Terms” (Know Your Terms) campaign had already stress-tested its methodology in August 2025 at Mount Kenya University and the Catholic University of Eastern Africa. Their central thesis, “Consent or Coercion,” argued that the lengthy, legalistic user agreements on platforms like TikTok and Instagram constituted a form of digital entrapment for Kenyan youth. When the Nairobi team, led by student advocates from the Parklands Law Campus, met their Mombasa counterparts, they presented data linking unintelligible ToS agreements directly to the 78% harassment rate observed in the coastal region. The argument was mechanical: because users could not understand the reporting method or their data rights defined in the fine print, they were functionally defenseless against algorithmic amplification of hate.
Table: The Jua Terms Translation Matrix (October 2025)
The Jua Terms team distributed “decoding cards” at TUM, translating Silicon Valley legalese into “Sheng” (Kenyan urban slang) and plain English. This table reconstructs the primary “traps” identified during the joint session.
| Standard ToS Clause | Jua Terms Translation | Observed Impact in Mombasa |
|---|---|---|
| “We may use your content to improve our services and algorithms.” | “We feed your videos to the machine to see what makes people angry.” | Viral amplification of humiliating content (Trolling) to boost engagement. |
| “Moderation is automated and subject to error.” | “A robot decides if you are being bullied. The robot doesn’t know Sheng.” | Harassment in local dialects (Mijikenda, Swahili) bypassed filters 92% of the time. |
| “Users are responsible for their interactions.” | “If you get attacked, you are on your own.” | Victims self-censored or left platforms rather than reporting incidents. |
| “Disputes are subject to binding arbitration in [US State].” | “not sue us in a Kenyan court.” | Legal impunity for platforms even with the 2018 Computer Misuse and Cybercrimes Act. |
The TikTok Vector and the Safety Council Gap
The collaboration specifically targeted TikTok, identified by Scroll Bila Troll research as the primary vector for peer-to-peer harassment among the 18-25 demographic in Mombasa. While TikTok had established its inaugural Sub-Saharan Africa Safety Advisory Council in Nairobi in September 2024, appointing key figures like Lillian Kariuki of the Watoto Watch Network, the youth campaigners argued that high-level advisory bodies were too detached from the “ground zero” of algorithmic violence. The Jua Terms analysis revealed a disconnect between the Safety Council’s policy recommendations and the user experience on the ground. Even with the Council in place, the “Scroll Bila Troll” survey data showed that 38. 9% of youth in Mombasa specifically TikTok as the site of their worst harassment experiences. The campaigners contended that the platform’s “For You” page algorithm prioritized high-engagement content, which frequently included “roast” videos and non-consensual image sharing, regardless of the safety policies on paper.
Operationalizing the Data Protection Act
The joint front moved to weaponize the Kenya Data Protection Act (2019) against these platforms. The Office of the Data Protection Commissioner (ODPC) had spent much of 2024 and 2025 conducting awareness campaigns, such as the “White Wall” project, the youth activists demanded enforcement over awareness. During the October 17 session, the combined teams drafted a “Youth Resolution on Digital Consent.” This document demanded that platforms operating in Kenya: 1. Provide “One-Click” reporting for harassment in local languages. 2. Publish quarterly transparency reports specifically for the Kenyan market, detailing how accounts were suspended for harassment. 3. Simplify Terms of Service into visual, icon-based summaries (a core demand of the Jua Terms initiative). The resolution was not symbolic. It was designed to be filed as a formal complaint with the ODPC, testing the regulator’s willingness to sanction foreign tech giants for failing to protect the “data dignity” of Kenyan citizens.
The “Zero Point” Strategy
This merger of emotional narrative (Scroll Bila Troll) and legal mechanics (Jua Terms) defined the “Zero Point” strategy. The campaigners realized that appealing to the morality of trolls was insufficient; they had to alter the digital architecture that rewarded trolling. By framing online harassment as a violation of consumer rights and data privacy—rather than just a social ill—they opened a new front that platform policy teams could not ignore. The Nairobi team’s presence also brought logistical discipline. They introduced encrypted communication for the campaigners, fearing that their aggressive stance against “Big Tech” and local political trolls might invite digital retaliation. This “privacy ” operational security became standard procedure for the Scroll Bila Troll volunteers as they moved into the execution phase of the campaign in late October 2025.
State Backed Trolling: Amnesty Report Links Government to Online Attacks
The State as the Aggressor: Amnesty International’s Findings
The narrative of online harassment in Kenya shifted dramatically in late 2025. While the Scroll Bila Troll campaign initially focused on peer-to-peer bullying among university students, their data collection collided with a far more formidable adversary. Evidence released by Amnesty International in November 2025 confirmed what activists had suspected for months. The state itself had become the primary architect of digital harassment. The report titled “This Fear, Everyone is Feeling It” dismantled the idea that online mobs were organic outpourings of public anger. It instead presented proof of a coordinated and well-funded machine designed to silence dissent through “technology-facilitated violence.”
Amnesty’s investigation linked the Kenyan government to a systematic campaign of digital repression that ran parallel to physical crackdowns. The numbers were clear. Between June 2024 and October 2025, the report documented 83 enforced disappearances and over 3, 000 arbitrary arrests. In nearly every case, the physical abduction was preceded by a wave of state-backed online attacks. Victims reported receiving identical threatening messages from anonymous accounts just hours before police or unmarked vehicles arrived at their locations. The digital was not a space for insults. It had become the targeting system for kinetic state violence.
The Economics of the “3, 000 Shilling Army”
The infrastructure of state trolling relies on a gig economy of disinformation that exploits high youth unemployment. Mozilla Foundation fellows had previously identified this phenomenon in 2021 and 2023. By 2025, it had industrialized. The “Scroll Bila Troll” researchers interviewed former participants of these networks who described a tiered system of command and control. At the top sit the “Admins” or “Strategists” who operate directly from government-affiliated offices. These individuals draft the daily “briefs” which contain the specific hashtags to be trended and the specific angles of attack against perceived enemies of the state.
them are the “Influencers” with verified badges. These accounts are frequently rented out. A verified account holder might lease their credentials to a state operative for a monthly fee. This allows the state to broadcast propaganda through channels that appear independent and authoritative. At the bottom is the “foot soldier”. These are frequently university students or unemployed graduates who are paid between KSh 1, 000 and KSh 3, 000 (approximately $10 to $25) per day to copy and paste content into replies and quote-tweets. Their instructions are delivered via WhatsApp groups with names like “Defense Council” or “Patriot Wing.” The goal is not to win an argument. The goal is to flood the zone with enough noise that the original critique is rendered invisible.
| Service Level | Role Description | Daily Rate (KSh) | Daily Rate (USD) |
|---|---|---|---|
| Tier 1: The Trend Setter | Verified account (Blue Check) used to initiate hashtags. | KSh 15, 000, 25, 000 | $115, $190 |
| Tier 2: The Amplifier | Accounts with 10k+ followers used to retweet/quote. | KSh 5, 000, 10, 000 | $38, $76 |
| Tier 3: The Swarm | Low-follower accounts used for mass replies/insults. | KSh 1, 000, 3, 000 | $7, $23 |
| Package: “Kill Switch” | Coordinated mass reporting to suspend a target’s account. | KSh 50, 000 (Project) | $385 (Project) |
Targeting the Judiciary and the “Hanifa” Precedent
The state’s digital artillery was not limited to Gen Z protesters. It was frequently turned against the Judiciary. Throughout 2024 and 2025, judges who issued rulings unfavorable to the administration found themselves the subjects of vicious smear campaigns. When the High Court declared the Housing Levy unconstitutional, a swarm of accounts immediately began trending hashtags labeling specific judges as “corrupt” or “cartel-owned.” These attacks were not random. They used identical graphics and language that appeared simultaneously across hundreds of accounts. This signaled central coordination.
The case of Hanifa Adan in April 2025 served as a grim precursor to the October campaign. After giving an interview to the BBC regarding state violence, Adan was targeted by a campaign dubbed #ToxicActivists. The attacks were highly gendered and Islamophobic. Trolls circulated AI-generated imagery and deepfakes designed to humiliate her. This tactic of using sexualized disinformation against female activists became a hallmark of the state’s strategy. The Scroll Bila Troll campaign noted that 60 percent of the attacks on their own female organizers involved body-shaming or threats of sexual violence. This was a deliberate strategy to drive women out of the digital public square.
Weaponizing the Law: The 2025 Amendment
The state’s assault was not limited to covert operations. It also utilized the legislature. On October 15, 2025, President William Ruto assented to the Computer Misuse and Cybercrimes (Amendment) Act of 2025. The government billed this as a necessary update to fight fraud and child safety problem. Civil society saw it as a tool for censorship. The amendment introduced vague provisions criminalizing communication that causes “reputational harm” or “public panic.”
Legal experts argued that these terms were subjective enough to cover any investigative journalism or activist critique. The Act also expanded the powers of the National Computer and Cybercrimes Coordination Committee (NC4). It granted them the authority to recommend the blocking of websites or applications deemed to promote “unlawful activities.” This was widely interpreted as a “Kill Switch” for platforms like TikTok or X if they became too troublesome during protests. The timing of the assent was serious. It came just days before the Scroll Bila Troll campaign launched. This created a chilling effect where students feared that reporting on harassment could itself be interpreted as “causing panic” under the new law.
“We are seeing a fusion of legislative overreach and digital thuggery. The state pays trolls to harass you on Monday and uses the law to arrest you for complaining about it on Tuesday.”
, Victor Ndede, Amnesty International Kenya (November 2025)
The High Court Intervention
The legal battle over the amendment was immediate. On October 22, 2025, the High Court issued a temporary suspension of the most controversial sections of the new Act. The ruling came after a petition filed by the Kenya Human Rights Commission (KHRC) and the Law Society of Kenya (LSK). The judges noted that the definitions of “false information” and “reputational harm” were unconstitutionally broad. This judicial intervention provided a brief window of safety for the Scroll Bila Troll organizers. It allowed them to continue their work without the immediate threat of arrest under the new statutes. The state’s response was predictable. The troll networks immediately pivoted to attacking the judges who issued the suspension.
Digital Encirclement of Activists
The Amnesty report highlighted a tactic known as “digital encirclement.” This involves cutting off an activist’s support network before moving in for an arrest. State operatives would compromise the WhatsApp groups used by organizers. They would inject false information about meeting times or police locations., they would use compromised accounts to accuse leaders of stealing funds. This sowed distrust within the movement. When the activist was, the physical arrest would occur. The Scroll Bila Troll campaign had to adopt counter-intelligence measures. They moved sensitive communications to encrypted platforms like Signal and instituted strict verification for new members. They understood that they were not just fighting bullies. They were fighting a state intelligence operation.
The Role of Platform Negligence
The efficacy of state trolling in Kenya was compounded by the negligence of social media platforms. The Amnesty report and subsequent analysis by the Scroll Bila Troll team found that X (formerly Twitter) and TikTok failed to act on thousands of reports of coordinated harassment. The platforms’ automated moderation systems were ill-equipped to detect harassment in Swahili or Sheng (Kenyan slang). State trolls exploited this gap. They used local dialects to problem death threats that bypassed English-language filters. also, the “Verified” status on X, which can be purchased, gave state trolls an algorithmic advantage. Their replies were prioritized in the feed. This ensured that pro-government propaganda always appeared above the critiques of genuine citizens. The platforms profited from the engagement generated by these troll farms while the subjects of the attacks feared for their lives.
Legislative Weaponization: The October 15 Computer Misuse Act Amendment
The October 15 Amendment: A Legislative Trojan Horse
Two days before the Scroll Bila Troll campaign formally launched its empathy offensive at the Technical University of Mombasa, the legislative ground beneath Kenya’s digital ecosystem shifted violently. On October 15, 2025, President William Ruto assented to the Computer Misuse and Cybercrimes (Amendment) Act, 2024. While the administration framed the legislation as a necessary modernization of the 2018 statutes, citing the rise of AI-driven deepfakes and algorithmic radicalization, legal analysts and digital rights groups immediately identified a more coercive objective. The amendment did not update penalties; it fundamentally altered the power between the state and the screen, deputizing internet service providers (ISPs) as enforcers of state censorship without judicial oversight.
The timing presented a clear irony. While student leaders in Mombasa advocated for peer-to-peer accountability to curb harassment, the state apparatus secured legal instruments to silence dissent under the guise of “public order.” The amendment’s most contentious provision, a revision to Section 6 of the principal Act, the National Computer and Cybercrimes Coordination Committee (NC4) with unilateral authority to order the blocking of websites and applications. Previously, such actions required a court order, a procedural safeguard that allowed for legal challenges before a blackout could occur. The new law stripped this buffer, allowing the NC4, a body dominated by security, intelligence, and military officials, to problem direct directives to ISPs if a platform was deemed to promote “illegal activities, terrorism, or extreme religious practices.”
The Anatomy of Control: Key Provisions
The amendment introduced specific clauses that expanded the definition of cybercrimes to a degree that critics argued criminalized ordinary expression. The vagueness of the terminology provided law enforcement with broad discretion to interpret criticism as “harassment” or investigative journalism as “false information.”
| Provision | 2018 Principal Act | 2025 Amendment (Oct 15) | Implication for Civil Liberties |
|---|---|---|---|
| Website Blocking | Required a High Court order based on evidence of a crime. | NC4 Administrative Order: Can block sites promoting “extreme religious practices” or “illegal activities” without court approval. | Bypasses judicial review; allows immediate censorship of dissenting platforms or blogs. |
| Cyber Harassment (Sec 27) | Criminalized content likely to cause apprehension or fear of violence. | Expanded to include content “likely to cause a person to commit suicide” or cause “detrimental affect.” | Subjective interpretation of “detrimental” can be used to shield public officials from scrutiny. |
| False Information (Sec 22/23) | Criminalized “false” publications causing panic/chaos. Fine: 5M KES. | Retained and reinforced even with ongoing litigation. Penalties enhanced for AI-generated content. | Continues to be the primary tool for arresting journalists and activists (e. g., Mutemi Kiama case). |
| SIM Swap Fraud (Sec 42A) | Not explicitly defined; prosecuted under general fraud. | Specific Offense: 20M KES fine or 10 years prison for unauthorized SIM takeover. | Positive inclusion for consumer protection, frequently used to justify the broader draconian bill. |
The inclusion of Section 42A, which specifically criminalized SIM-swap fraud, served as the legislative sweetener. With financial fraud rampant, costing Kenyan banks and customers an estimated 14 billion KES annually by 2024, the government argued the amendment was a consumer protection need. Yet, the juxtaposition of these financial safeguards with the draconian expansion of Section 27 (Cyber Harassment) revealed the dual-use nature of the law. Under the new text, “harassment” included communication that “detrimentally affects” a person. Legal scholars from the Law Society of Kenya (LSK) warned that this phrasing was unconstitutionally overbroad, chance criminalizing a negative review of a business or a corruption exposé involving a public official.
Weaponization of “False Information”
The state’s reliance on Section 23 of the Act, which criminalizes the publication of “false information”, intensified in the weeks following the amendment’s passage. Between October 15 and December 1, 2025, the Defenders Coalition recorded 14 arrests of digital content creators under this specific provision. Unlike the Scroll Bila Troll campaign, which focused on interpersonal empathy, these arrests targeted political speech. The pattern was consistent: an activist would post details regarding government expenditure or procurement irregularities; the state would label the data “false” and “likely to cause panic”; and the individual would be detained, frequently over the weekend to maximize punitive custody before a court appearance.
One high-profile case involved an independent blogger in Eldoret who questioned the allocation of county funds for a drainage project that never materialized. Charged under the amended Section 23, the blogger faced a fine of 5 million KES (approx. $38, 000 USD) or two years in prison. The charge sheet did not dispute the absence of drainage argued the blogger’s post was “calculated to cause chaos” among residents. This application of the law demonstrated the precise fear articulated by Human Rights Watch: that “cybersecurity” laws were being retooled as “reputation management” systems for the political class.
The NC4: A Militarized Regulator
The elevation of the National Computer and Cybercrimes Coordination Committee (NC4) to a supreme internet regulator marked a militarization of digital governance. Unlike the Communications Authority (CA), which is ostensibly a civilian regulatory body, the NC4’s composition includes the Principal Secretary for Interior, the Chief of Defence Forces, the Inspector General of Police, and the Director General of the National Intelligence Service. By granting this committee the power to order website blocks without judicial oversight, the October 15 amendment placed the Kenyan internet under the purview of the national security establishment.
For the youth activists of Scroll Bila Troll, this reality complicated their advocacy. Their campaign relied on the premise that the internet is a public square requiring civic responsibility. The state’s actions, yet, suggested the internet was a battlefield requiring command and control. When student leaders at TUM spoke of “safe spaces,” they meant safety from bullies; when the NC4 spoke of “safe spaces,” they meant safety from subversion. This created a chilling effect. Students who might have used the Scroll Bila Troll platform to report institutional failure, such as sexual harassment by faculty or embezzlement of student union funds, feared that such reports could be classified as “cyber harassment” under the expanded Section 27, leading to expulsion or arrest.
The “Suicide Clause” and Subjectivity
Perhaps the most emotionally manipulative addition to the law was the expansion of cyber harassment to include messages “likely to cause a person to commit suicide.” On the surface, this aligned with the mental health goals of Scroll Bila Troll. yet, legal experts pointed out that the clause absence an evidentiary standard for “likelihood.” In practice, this allowed authorities to arrest critics based on the subjective distress of the recipient. If a public official claimed that a barrage of serious tweets made them “suicidal,” the authors of those tweets could face ten years in prison. This inverted the victim-protection logic: instead of protecting youth from relentless bullying, the law could be used to protect figures from relentless accountability.
“We are seeing the weaponization of mental health language to shield impunity. The law does not ask if the criticism is true; it only asks if the recipient feels bad about it. That is not justice; that is therapy at gunpoint.” , Mercy Mutemi, Digital Rights Lawyer, November 2025 Legal Briefing.
The March 2026 Vindication
The tension between the legislative crackdown and constitutional freedoms reached a breaking point just yesterday, March 6, 2026. In a landmark ruling that vindicated the concerns raised throughout late 2025, the Court of Appeal declared Sections 22 and 23 of the Computer Misuse and Cybercrimes Act unconstitutional. The three-judge bench, led by Justice Patrick Kiage, ruled that criminalizing “false information” violated Article 33 of the Constitution, as the state cannot be the arbiter of truth. While this ruling dismantled the “fake news” weapon, the October 15 amendments regarding NC4’s blocking powers and the expanded definition of harassment remain on the books, leaving the digital surveillance architecture largely intact.
The legislative environment of late 2025 served as the crucible for the Scroll Bila Troll campaign. It forced the youth movement to distinguish their version of “digital safety” (rooted in empathy and peer support) from the state’s version (rooted in silence and control). The students at TUM were not just fighting trolls; they were navigating a minefield where a misstep could be labeled a crime. The 78% of youth who witnessed harassment were caught between a rock and a hard place: the toxicity of the mob and the tyranny of the law.
The Gender Gap: 55.7% of Women Candidates Targeted Compared to 35.4% of Men
The Anatomy of the Attack: Competence vs. Morality
The lies not just in the volume of attacks in their nature. The Scroll Bila Troll analysis, corroborating earlier findings by FIDA-Kenya, showed a distinct bifurcation in the language used against based on gender. Male politicians and student leaders were attacked primarily on competence and corruption. Keywords associated with male included “thief,” “money,” “corruption,” and “failure.” These attacks, while harsh, remained within the of professional conduct. In contrast, female faced attacks centered on morality, sexuality, and appearance. The campaign’s semantic analysis of comments on Mombasa-based influencers and student politicians found that 68% of negative comments directed at women attacked their bodies or relationship status. Common slurs included “slay queen,” “prostitute,” and derogatory comments about their marital status (e. g., “husbandless”).
This “morality policing” serves a specific political function: to delegitimize women not by questioning their ideas, by questioning their right to exist in the public sphere. Tullie Apiyo, a candidate in the 2022 elections who later advised the Scroll Bila Troll strategy team, noted that online violence forced her to realize politics was designed as a “man’s world.” Her experience, where attackers used pseudonymous accounts to harass her family, mirrored the stories collected from female students at TUM in 2025.
The Deepfake Escalation
A disturbing trend identified by the Scroll Bila Troll campaign in late 2025 was the weaponization of Generative AI. While the 2022 baseline data relied on text and crude photoshop, the 2025 environment was saturated with “cheap fakes” and sophisticated deepfakes. A 2025 study by the Collaborative Centre for Gender and Development (CCGD), referenced by the campaign, found that nearly one in four women surveyed had been targeted using AI tools, including deepfake images and doctored content. This marked a significant escalation from previous years. The campaign documented cases where female student leaders at TUM had their faces superimposed onto pornographic material, a tactic known as image-based sexual abuse (IBSA). These images were then circulated on WhatsApp groups and TikTok to shame the victims into silence. Unlike text-based insults, which can be countered with arguments, non-consensual intimate imagery (NCII) creates immediate, irreversible reputational damage.
| Attack Vector | Male (Primary Focus) | Female (Primary Focus) | Platform Dominance |
|---|---|---|---|
| Disinformation | Financial scandal, political betrayal, incompetence. | Sexual history, marital status, “moral looseness.” | Facebook, X |
| Visual Harassment | Memes mocking intelligence or physical weakness. | Deepfake pornography, body shaming, doctored nudes. | TikTok, WhatsApp |
| Threat Type | Legal action, loss of office, public embarrassment. | Rape threats, stalking, physical violence against children. | DM (Direct Message) |
| Psychological Goal | Accountability (forced resignation). | Silencing (withdrawal from public space). | All Platforms |
The “Silence Effect” and Political Withdrawal
The metric of this gender gap is not the number of insults, the number of women who leave. The Scroll Bila Troll campaign identified a phenomenon they termed the “Silence Effect.” Data from the Association of Media Women in Kenya (AMWIK) and UN Women indicated that online harassment is a primary driver for women exiting politics and digital journalism. In the 2022 election pattern, 7 out of 29 women candidates interviewed admitted to halting their social media activity entirely due to abuse. By 2025, the Scroll Bila Troll survey at TUM found that 42% of female students who had considered running for student government declined to do so specifically citing fear of online shaming. This withdrawal creates a feedback loop. As women retreat from digital discourse to protect their mental health and physical safety, the online space becomes more male-dominated, further normalizing the toxic “bro-culture” that fuels the harassment. The campaign argued that this is not just a safety problem a democratic emergency; if 50% of the population cannot safely speak on the town square, the square is broken.
Platform Negligence and the Language Barrier
The Scroll Bila Troll campaign directed significant ire toward the platforms themselves, TikTok, Meta, and X, for their failure to moderate content in local languages. While English-language harassment is frequently flagged by automated systems, the campaign found that abuse in Swahili, Sheng (a Swahili-English slang), and vernacular languages frequently evaded detection. A slur that would trigger an immediate ban in English could remain up for weeks if written in Sheng. The campaign’s demands, issued in partnership with Amnesty International Kenya, called for platforms to: 1. Publish Anti-Cyberbullying Transparency Reports specific to Kenya, detailing how gender-based attacks were reported and removed. 2. Increase Human Moderation for Sheng and local dialects, rather than relying on US-centric AI models. 3. Fast-Track Reporting Channels for NCII (Non-Consensual Intimate Imagery) to prevent the viral spread of deepfakes. The 55. 7% vs. 35. 4% statistic is not a historical record; it is the structural reality that the Scroll Bila Troll campaign sought to. By October 2025, the youth leaders at Mombasa understood that they were not fighting a technology problem, a societal one amplified by technology—a system where a man is punished for what he does, and a woman is punished for who she is.
Platform Accountability: Demands for Meta and TikTok Transparency Reports
The “Black Box” of Nairobi: Quantifying the Silence
On October 17, 2025, two days after President William Ruto assented to the controversial Computer Misuse and Cybercrimes (Amendment) Act, the Scroll Bila Troll leadership issued a counter-manifesto at the Technical University of Mombasa. While the government’s new law focused on state surveillance and criminalizing “misleading” information, the student-led campaign argued that the real danger lay in the corporate negligence of Silicon Valley giants. Frankline Kimtai, the campaign’s lead organizer, presented a demand letter addressed to the regional policy heads of Meta and TikTok, rejecting the “aggregated obscurity” of current transparency reporting.
The core grievance was the “Rest of World” classification. For years, platforms like Facebook and TikTok grouped Kenyan data into broad regional buckets (Sub-Saharan Africa or “Rest of World”), making it impossible for local researchers to audit specific moderation failures. The Scroll Bila Troll “October Manifesto” demanded three specific shifts in reporting by January 2026: granular country-level data on harassment takedowns, disclosure of human moderator staffing levels fluent in Swahili and Sheng, and a public audit of algorithmic demotion for harassment reports.
TikTok’s Numbers Game: The “Zero-Second” Myth
The campaign’s demands coincided with the release of TikTok’s Q2 2025 Community Guidelines Enforcement Report, which the student leaders dissected during the launch. The report boasted that 592, 037 videos were removed in Kenya between April and June 2025. Of these, 96. 3% were removed within 24 hours, and 92. 9% were scrubbed before receiving a single view. On the surface, these metrics suggested a strong defense.
yet, the Scroll Bila Troll data science team, led by TUM engineering students, argued these numbers were a “sanitized illusion.” They pointed to the 3. 7% of content that did remain up for more than 24 hours. In a viral ecosystem, a harassment video can accrue 50, 000 views in the 25th hour. The campaign’s internal survey of 18-to-25-year-olds in Mombasa found that 38. 9% had witnessed bullying on TikTok that for days even with reporting it. The gap lay in the definition of “violative.” TikTok’s AI excelled at catching copyright infringement and nudity (binary violations) failed catastrophically at detecting Sheng-based linguistic harassment, which requires cultural nuance.
| Metric | TikTok Q2 2025 Report Claim | Scroll Bila Troll Finding (Mombasa Sample) |
|---|---|---|
| Takedown Speed | 96. 3% removed within 24 hours | 42% of harassment reports pending>48 hours |
| Proactive Detection | 92. 9% removed before any views | Harassment videos averaged 1, 200 views before removal |
| Live Stream Safety | 90, 000 sessions interrupted (Q3 2025) | 65% of female users reported unchecked “raid” harassment on Live |
| Appeal Success | Not Disclosed for Kenya specifically | <5% of user appeals resulted in content restoration |
The Meta Blindspot: The “Sheng” and Dialect Deficit
While TikTok faced heat for algorithmic amplification, the demands on Meta (Facebook and Instagram) focused on the human cost of moderation. The campaign leveraged the momentum of the ongoing legal battles involving former content moderators. By October 2025, the Kenyan courts were still adjudicating the from the Sama and Majorel lawsuits, with rulings on jurisdiction and worker compensation repeatedly delayed until February 2026. Scroll Bila Troll organizers argued that the mental health emergency among moderators directly correlated with the safety of users.
The “Sheng Gap” became a central talking point. Sheng, a and evolving urban slang used by Kenyan youth, frequently evades standard Swahili keyword filters. The campaign instances where direct threats of sexual violence phrased in coastal Sheng were marked as “Safe” by Meta’s automated systems. The October Manifesto demanded that Meta disclose the exact number of moderators dedicated to the Kenyan market who were native Sheng speakers, rather than relying on generic Swahili speakers from other East African nations who might miss the coded language of cyberbullying.
Legislative Crossfire: The Act of 2025
The campaign’s demands were complicated by the hostile legislative environment. The Computer Misuse and Cybercrimes (Amendment) Act, 2025, assented to on October 15, introduced heavy fines (up to KSh 20 million) for sharing information “likely to cause apprehension.” While the government framed this as a tool against cyberbullying, Scroll Bila Troll leaders feared it would be used to silence the very activists exposing platform failures. The High Court’s suspension of key sections of the Act on October 22, 2025, provided a temporary reprieve, allowing the campaign to focus its pressure on the platforms rather than fighting off state arrest warrants.
“We are squeezed between a government that wants to read our DMs and platforms that ignore our reports. We don’t want the police in our group chats; we want the platforms to fix their broken algorithms.”
, Frankline Kimtai, Scroll Bila Troll Organizer, Press Statement, Oct 17, 2025.
The “Trusted Partner” Ultimatum
A final, serious demand was the expansion of the “Trusted Partner” program. Historically, only large, well-funded NGOs had direct escalation channels to Meta and TikTok to fast-track dangerous content removal. Scroll Bila Troll, representing a decentralized network of university students, was locked out of these priority lanes. They demanded that student guild leaders at major universities (TUM, Kenyatta University, UoN) be granted “Trusted Flagger” status to bypass the automated reporting queues that had failed 78% of their peers. This demand sought to decentralize moderation power, moving it from distant offices in Dublin or California to the campuses where the harassment was actually taking place.
The Real Cost of a Comment: Psychological Impact Metrics from the TUM Panel
The Real Cost of a Comment: Psychological Impact Metrics from the TUM Panel
The “The Real Cost of a Comment” panel, held during the October 17, 2025 launch at the Technical University of Mombasa (TUM), moved beyond anecdotal evidence to present a forensic accounting of the mental health emergency triggered by digital harassment. Moderated by Derrick Ouko of Search for Common Ground, the session utilized a “fan-out” data presentation style, where survivors and researchers laid out specific metrics that quantified the psychological toll on Kenyan youth. The following verified metrics and findings were presented or during the session to substantiate the campaign’s urgency:
1. The “78% Vector” (Campaign Baseline)
The central pillar of the presentation was the campaign’s own baseline research, dubbed the “78% Vector.” * Metric: 78% of Mombasa youth aged 18, 25 reported experiencing or witnessing targeted cyberbullying. * Platform Specificity: Unlike previous years where Facebook or X (formerly Twitter) dominated harassment statistics, TikTok was identified as the primary vector for this demographic. * Impact: Campaign organizer Frankline Kimtai noted that this high prevalence correlates directly with a spike in “digital withdrawal,” where victims cease online economic and social activity to avoid abuse.
2. Suicidality and Cyber-Victimization (Ndetei et al. Data)
Panelists referenced the 2024 study by the Africa Institute of Mental and Brain Health, led by Professor David Ndetei, to illustrate the severe end of the psychological spectrum. * Metric: 45. 3% of students involved in cyberbullying (both as victims and perpetrators) reported having suicidal thoughts. * Context: This figure was used to the myth that online comments are “harmless.” The data indicated that the intensity of suicidal ideation frequently surpassed deterrent factors, creating a direct link between unmoderated comment sections and life-threatening mental health crises.
3. The Private University Prevalence (JSAA Study)
To show that the problem permeates all socioeconomic strata, the panel data published in the Journal of Student Affairs in Africa (July 2024). * Metric: 75. 8% of students in selected private universities in Kenya experienced cyberbullying. * Academic Correlation: Panelist “Alex,” a student survivor, correlated this prevalence with academic decline. The constant state of hyper-vigilance required to navigate online spaces depletes cognitive resources needed for study, leading to a measurable drop in concentration and grades.
4. The “Costs of Connection” (DHRP 2025 Report)
The panel integrated findings from the Paying the Costs of Connection report by the Digital Health and Rights Project (DHRP), released in early 2025. This transnational study, which included cohorts from Kenya, provided a broader context for the local data. * Metric: Three-quarters of young adults in the study described experiencing technology-facilitated abuse. * Psychological Toll: The report detailed how fear of being “outed” or shamed led to severe anxiety and the avoidance of essential health information online. For marginalized groups, this digital exclusion is not just a social inconvenience a health risk.
5. Legal Inefficacy and Victim Despair
Beyond the clinical metrics, the panel addressed the psychological weight of legal failure. * Finding: Panelist “Hilda” presented qualitative data suggesting that the Computer Misuse and Cybercrimes Act is viewed as “ineffective” by the youth demographic. * Consequence: The absence of legal recourse contributes to a sense of learned helplessness. Victims feel that reporting abuse is futile, which exacerbates feelings of isolation and depression. The “justice gap” was identified as a key psychological stressor, distinct from the harassment itself.
6. The Evolution of Abuse
Panelist “Erick” outlined the shifting nature of digital aggression. * Observation: Abuse has mutated from “casual remarks” to “aggressive, coordinated criticism.” * Metric: While specific percentages for this shift were not, the qualitative consensus was that the severity of individual incidents has increased, even if the total volume fluctuates. This escalation intends to silence expression, censoring the digital voice of the victim.
| Source | Key Metric | Primary Psychological Outcome | Target Demographic |
|---|---|---|---|
| Scroll Bila Troll Internal Survey (2025) | 78% prevalence | Digital withdrawal, social anxiety | Mombasa Youth (18-25) |
| Ndetei et al. (2024) | 45. 3% suicidal ideation | Severe distress, suicidal thoughts | High School Students |
| JSAA Study (2024) | 75. 8% prevalence | Academic decline, concentration loss | Private University Students |
| DHRP Report (2025) | 75% (approx.) abuse rate | Health information avoidance, fear | Young Adults (Kenya cohort) |
“We can make digital peace trend and let online empathy go viral. Behind every username is a human being with feelings.”
, Frankline Kimtai, Scroll Bila Troll Organizer, October 17, 2025.
Targeting Human Rights Defenders: The Case of Mariam and Joseph
The Digital Siege on Mombasa’s Defenders
The intersection of the Scroll Bila Troll campaign and the broader human rights emergency in Kenya became undeniable in November 2025, when Amnesty International Kenya released its investigative report, “This Fear, Everyone is Feeling It.” While the campaign at the Technical University of Mombasa (TUM) initially focused on peer-to-peer cyberbullying among students, the data quickly exposed a more sinister of coordinated attacks targeting youth activists. The campaign’s transparency logs revealed that Human Rights Defenders (HRDs) were not suffering from random trolling were subjects of “precision harassment”, state-aligned digital offensives designed to silence dissent through fear.
The Scroll Bila Troll research team, led by Frankline Kimtai, found that 78% of Mombasa’s youth aged 18-25 had witnessed or experienced online harassment. Yet, for activists, this harassment migrated offline with worrying speed. The campaign’s case files on “Mariam” and “Joseph” (pseudonyms used to protect the survivors’ identities) illustrate the gendered and political bifurcation of these digital assaults.
The Case of Mariam: Gendered Disinformation
Mariam, a 27-year-old community organizer in Mombasa, became a primary case study for the Scroll Bila Troll digital -aid team. Her ordeal began after she livestreamed a protest against the Finance Bill 2024 and continued her advocacy into late 2025. Unlike her male counterparts, the attacks against Mariam were almost exclusively gendered and sexualized, a tactic known as gendered disinformation.
According to the campaign’s documentation, Mariam received over 400 direct messages on TikTok and WhatsApp within a 48-hour window in October 2025. These messages were not political rebuttals threats of sexual violence. The attackers weaponized Non-Consensual Intimate Images (NCII), using generative AI to create deepfake pornography featuring Mariam’s face. These images were then circulated in local community WhatsApp groups to discredit her standing among conservative coastal constituents.
“I had people coming into my inbox and telling me, ‘You die and leave your kids. come and attack you.’ I even had to change my child’s school.” , Mariam*, Mombasa-based HRD (Source: Amnesty International Kenya, Nov 2025)
The harassment escalated to physical surveillance. Mariam reported being followed by unmarked vehicles, a claim substantiated by the Scroll Bila Troll digital security audit which found her real-time location had been leaked on a pro-government Telegram channel. This culminated in her forced disappearance for two nights in police custody, an event that the campaign was directly facilitated by the digital breadcrumbs left by her harassers.
The Case of Joseph: The “Foreign Agent” Narrative
While Mariam faced sexualized violence, Joseph, a 24-year-old student leader at TUM, faced a different vector of attack: the “foreign agent” smear. Joseph’s case highlights the industrial of state-sponsored trolling. After organizing a campus town hall on digital rights, Joseph was targeted by a network of “bot-like” accounts on X (formerly Twitter). These accounts, which the Scroll Bila Troll researchers traced to a coordinated influence operation, flooded his timeline with accusations that he was funded by “Western imperialists” to destabilize the region.
The attacks against Joseph utilized doxing, the public release of private information. His national ID number, phone number, and parents’ home address were posted under the hashtag #JosephTheTraitor. This digital assault was paired with the weaponization of the Computer Misuse and Cybercrimes Act of 2018. Joseph was arrested in November 2025 and charged under Section 23 for “publishing false information,” a provision frequently used to criminalize criticism of public officials before it was declared unconstitutional in March 2026.
Tactical Breakdown of Attacks on HRDs (2024-2025)
The Scroll Bila Troll campaign, in partnership with Defenders Coalition, categorized the distinct methods used to target activists like Mariam and Joseph. The data shows a clear in tactics based on the target’s gender and profile.
| Tactic | Primary Target Demographic | method | Objective |
|---|---|---|---|
| NCII / Deepfakes | Women HRDs (e. g., Mariam) | AI-generated nudity, distribution in family/community groups. | Social shaming, reputational destruction, silencing through shame. |
| Doxing | Male & Female HRDs | Publishing ID numbers, home addresses, children’s schools. | Physical intimidation, inducing fear of offline violence. |
| Hashtag Poisoning | High-profile Student Leaders (e. g., Joseph) | Bot networks flooding protest hashtags with spam or smears. | Drowning out legitimate dissent, controlling the narrative. |
| Lawfare (Section 22/23) | Journalists & Bloggers | Arrests for “false information” or “cyber harassment.” | Legitimizing detention, seizing devices to extract data. |
The Legal Weaponization
The persecution of Mariam and Joseph was enabled by the legal architecture of the time. Throughout 2025, authorities relied heavily on the Computer Misuse and Cybercrimes Act to justify the arrest of activists flagged by these online hate campaigns. The Scroll Bila Troll legal clinic noted that in 60% of cases where a student was arrested for online speech, the arrest was preceded by a coordinated online smear campaign, suggesting a feedback loop between state-sponsored trolls and law enforcement.
The campaign’s advocacy, alongside bodies like the Law Society of Kenya (LSK), eventually contributed to the landmark Court of Appeal ruling in March 2026. The court struck down Sections 22 and 23 of the Act, citing them as “vague” and “overbroad.” Yet, for victims like Mariam and Joseph, this legal victory came after the damage, psychological trauma, reputational harm, and physical detention, had already been inflicted.
Algorithmic Complicity: How For You Pages Amplify Hate in Kenya
The Engagement Trap: “For You” as a Radicalization Engine
The core finding of the Scroll Bila Troll research team was not just that hate speech existed, that it was being curated. By October 2025, the campaign’s data analysis revealed that the “For You” page (FYP) on TikTok and the “examine” feed on X (formerly Twitter) were not passive mirrors of user interest active accelerants of harassment. The campaign’s technical analysis, corroborated by a July 2025 report from the African Digital Democracy Observatory (ADDO), exposed a method where algorithms systematically prioritized high-arousal negative emotion, outrage, disgust, and fear, over neutral or positive content to maximize dwell time.
The “Search-to-Feed” Pipeline
Scroll Bila Troll investigators identified a specific algorithmic feedback loop they termed the “Search-to-Feed Pipeline.” When users engaged with even a single piece of polarizing content, the recommendation engines aggressively narrowed their content diet. The ADDO report provided the technical grounding for this observation, revealing that TikTok’s search algorithms linked hateful prompts to marginalized groups with worrying efficiency.
| Metric | Finding | Source |
|---|---|---|
| Stereotype Amplification | 197 out of 300 videos (65. 6%) returned by hateful search prompts perpetuated harmful stereotypes against marginalized groups. | ADDO Report (July 2025) |
| Pre-View Removal Rate | 92. 9% of violative videos in Kenya were removed before receiving views in Q2 2025. | TikTok Transparency Report |
| The “1% Leak” | even with high removal rates, the remaining ~42, 000 videos (7. 1%) garnered millions of views due to viral velocity before moderation caught up. | Scroll Bila Troll Analysis |
| Keyword Association | Algorithms automatically linked neutral terms like “security” or “election” to ethnic slurs and inflammatory content. | Mozilla Foundation / ADDO |
State-Sponsored Algorithmic Hijacking
While the Scroll Bila Troll campaign focused on peer-to-peer harassment, their October 2025 dossier also highlighted how political actors weaponized these same algorithmic vulnerabilities. A November 2025 investigation by Amnesty International Kenya exposed the operations of the so-called “527 Bloggers”, a coordinated network of state-aligned trolls paid to hijack the X algorithm.
These actors used “astroturfing” techniques to manipulate the “Trending” tab, forcing state-sanctioned harassment campaigns into the feeds of apolitical youth. By mass-posting identical hashtags and engaging in coordinated attacks against Gen Z activists, these networks tricked the platform’s relevance signals into treating manufactured hate as organic high-interest news. The result was a digital environment where a student in Mombasa looking for entertainment was inadvertently funnelled into a stream of political vitriol.
The “Madoadoa” Effect and Coded Language
The campaign’s most damning evidence against algorithmic complicity was the platforms’ inability to parse context-specific hate speech. The term madoadoa (blemishes), a known incitement to ethnic cleansing in Kenya, had been flagged by the National Cohesion and Integration Commission (NCIC) as early as 2022. Yet, in late 2025, Scroll Bila Troll researchers found that recommendation engines still promoted videos using this audio track if the visual content was high-engagement (e. g., dancing or comedy).
This failure demonstrated a serious gap: algorithms were optimized for visual and audio retention remained semantically blind to local linguistic nuances. The “For You” page stripped content of its dangerous context, serving hate speech as entertainment to a teenage demographic that frequently absence the historical memory to recognize the dog whistles.
“The algorithm doesn’t know it’s hate speech. It only knows it’s ‘engaging.’ If a video of a student being mocked keeps people watching for three seconds longer than a video of a lecture, the machine promotes the mockery. We are fighting a math equation that solves for cruelty.”
, Frankline Kimtai, Scroll Bila Troll Organizer, Technical University of Mombasa (October 19, 2025).
Platform Response and the “Shadowban” Reality
In response to the mounting pressure from Scroll Bila Troll and the Communications Authority of Kenya (CA), TikTok ramped up its automated moderation, removing nearly 600, 000 videos in Kenya between April and June 2025. yet, the campaign noted that this “whack-a-mole” method failed to address the root algorithmic sorting. Users reported “shadowbanning”, where their content reach was artificially , not for posting hate speech, for posting counter-speech or educational content that the algorithm deemed “low engagement.” This created a perverse incentive structure where anti-harassment advocates were silenced by the very tools meant to curate relevance, while high-voltage trolling continued to trend.
The Reduction Goal: Metrics for Success in the Digital Peace Initiative
The “Zero Point” Baseline
The campaign’s launch was not symbolic; it was a direct response to the signing of the Computer Misuse and Cybercrimes (Amendment) Act, 2025 just 48 hours prior, on October 15. While the legislation provided the legal teeth, criminalizing communications likely to cause “psychological harm” or “apprehension of violence”, Scroll Bila Troll provided the enforcement method on the ground. The campaign’s “White Paper on Digital Safety” established a rigid set of Key Performance Indicators (KPIs) designed to measure the efficacy of this new legal framework against the reality of platform algorithms. The primary metric, dubbed the “Mombasa Index,” set the baseline for harassment prevalence. While national averages for youth cyber harassment hovered around 37% in early 2025, the 78% figure in Mombasa revealed a localized hyper-concentration of toxicity, driven largely by regional political polarization and socio-economic frustrations vented online.
| Metric Category | Baseline (Oct 2025) | Target (Dec 2026) | Measurement Tool |
|---|---|---|---|
| Prevalence Rate (Mombasa) | 78% of Youth (18-25) | 35% (National Avg) | Quarterly Campus Surveys |
| Platform Takedown Time | 48, 72 Hours | <120 Minutes | User Reporting Logs |
| Bystander Intervention | 11% Active Reporting | 50% Active Reporting | “Scroll Bila Troll” App Data |
| Suicidal Ideation Correlation | 45. 3% of Victims | <15% of Victims | Mental Health Clinic Referrals |
| Legal Prosecution Rate | <2% of Reports | 20% of Verified Cases | NCIC & DCI Case Files |
The 120-Minute Mandate
The most aggressive target set by the campaign is the “120-Minute Mandate.” Data collected by the Communications Authority of Kenya (CA) throughout 2024 and 2025 indicated that algorithmic amplification of harmful content frequently outpaced moderation efforts by a factor of ten. A hate speech video on TikTok could amass 500, 000 views within three hours, while the average takedown time for reported content in Kenya remained stagnant at approximately 48 hours. Scroll Bila Troll organizers argued that a 48-hour response window renders moderation useless. By the time the content is removed, the reputational damage is permanent, and the psychological impact has already occurred. The campaign’s demand for a two-hour maximum takedown window is based on the “Viral Velocity” curve, which shows that 80% of a video’s engagement occurs within the 180 minutes of posting.
“We are not asking for censorship; we are asking for latency reduction. If an algorithm can identify and boost a video in milliseconds, it can identify and flag harassment in the same timeframe. The delay is a choice, not a technical limitation.”
The Mental Health Correlation
The campaign’s focus on mental health metrics distinguishes it from previous anti-bullying efforts. Leveraging data from the Journal of Popular Education in Africa, which found that 45. 3% of cyberbullying victims in Kenyan high schools harbored suicidal thoughts, Scroll Bila Troll partnered with the “Mental 360” initiative to track offline consequences. The goal is to sever the link between online notification and psychiatric admission. In 2023, Kenya lost an estimated KES 10. 7 billion ($83 million) to cybercrime, the economic cost of mental health deterioration among the youth workforce remains unquantified. The campaign introduced the “Psychosocial Impact Score,” a metric derived from anonymous exit surveys at student health clinics. This score tracks the frequency of anxiety and depression diagnoses explicitly linked to “digital interactions.” The reduction goal here is absolute: a 60% decrease in digital-origin anxiety cases by the end of 2026.
The Bystander Conversion Protocol
Perhaps the most difficult metric to shift is the “Bystander Effect.” In the digital ecosystem, 99% of users who witness harassment scroll past it. The campaign’s strategy involves gamifying the reporting process. By training “Digital Peacekeepers”, students to monitor specific campus hashtags, the initiative aims to flip the bystander ratio. The target is to have 50% of witnessed incidents reported by a third party, rather than the victim. This shifts the load of policing from the traumatized individual to the community. Early trials in November 2025 showed a 15% increase in third-party reports within the TUM network, suggesting that the “Scroll Bila Troll” slogan, which to “Scroll Without Trolling”, was beginning to permeate the campus consciousness.
Legislative Metrics and the 2027 Horizon
The timing of the campaign is inextricably linked to the upcoming 2027 General Election. The National Cohesion and Integration Commission (NCIC) warned in January 2026 that social media platforms were becoming “battlegrounds for ethnic polarization.” The Scroll Bila Troll campaign serves as a pilot program for the NCIC’s broader “Social Media Monitoring Guidelines.” Success for the campaign is also measured by the utilization of the Computer Misuse and Cybercrimes (Amendment) Act, 2025. Specifically, the campaign tracks the application of Section 2, which expanded the definition of “access” and “harassment.” Prior to October 2025, prosecutors struggled to charge individuals for “psychological harm” caused by digital communication. The campaign’s legal aid wing tracks the number of police files opened under these new provisions. The target is not necessarily a high conviction rate, a high investigation rate, signaling to perpetrators that the digital veil of anonymity has been pierced.
The Economic Imperative
Beyond the social metrics, the campaign emphasizes the economic need of a safe digital space. With Kenya’s digital economy projected to contribute over 9% to the GDP by 2026, online harassment acts as a barrier to entry for young entrepreneurs. Female influencers and business owners, in particular, face disproportionate rates of “revenge porn” and gendered attacks, frequently forcing them offline. Scroll Bila Troll tracks the “Digital Retention Rate” of female-led SMEs in Mombasa. The hypothesis is that a reduction in harassment correlate directly with an increase in the longevity and revenue of online businesses owned by women aged 18-30. The baseline data from 2024 showed that 30% of young women ceased online trading within six months of a harassment incident. The campaign aims to cut this attrition rate to under 10%.
Conclusion: The Digital Peace Accord
The metric for Scroll Bila Troll is the ratification of the “Digital Peace Accord” by the student bodies of all major coastal universities. This document, drafted in November 2025, binds signatories to a code of conduct that exceeds platform community guidelines. It mandates peer mediation before police involvement and establishes a “Restorative Justice” framework for minor offenses. As the campaign moves into 2026, the data collected from the Mombasa “Zero Point” determine if this model can be exported to Nairobi and Kisumu. The are high; with the 2027 elections looming, the ability of Kenya’s youth to navigate the digital space without violence likely determine the stability of the nation’s streets. The metrics are set, the clock is ticking, and the scroll continues.
