Generative AI, Social Media Disinformation, and Polarization During the 2024 Gen Z Protests in Kenya
Hillary BOBO. “Generative AI, Social Media Disinformation, and Polarization During the 2024 Gen Z Protests in Kenya” Mambo! No. 113, 2026.
| The central research question of the study was to examine the intersection between generative AI (Artificial Intelligence), youth-led movements, social media disinformation, and societal polarisation during the 2024 Gen Z protests through a mixed-method paradigm, focusing on social media content analysis, key informant interviews, and an online survey. The 2024 and 2025 Gen Z protests in Kenya were interpreted as spontaneous and amorphous digital unrest, mainly propelled by disinformation, making the protests seen as a security threat while overlooking the unfavourable socio-economic and political conditions that made the protests resonant with the young people. The protests had been sparked by government-proposed tax measures that the Kenyan populace, in particular the Gen Z found punitive. Artificial Intelligence, especially generative AI was used to produce and disseminate a significant amount of the content created at unprecedented scale and speed, that transformed digital activism. However, it also brought the dark side of using artificial intelligence in creating deepfakes and manipulated visuals, leading to widespread social media disinformation by both protesters and pro-government actors. Findings from the survey projected that 52.2 % of the survey respondents, with a majority being young people at 79.4%, agreed that AI-generated content had a significant influence in shaping public perception negatively. Findings also revealed that information exchanges happened mainly through TikTok at 36%, and X at 21%, while 27% used more than two social media platforms to consume AI-generated messages. These revelations show the close intersection between generative AI, youth-led movements, social media disinformation, and societal polarisation. |
Introduction
The advent of generative AI in the last few years has revolutionised and transformed global information ecosystems through the introduction of highly advanced tools that have enabled a fast and efficient way of content production and circulation (Kariuki & Gichanga, 2025). The 2024/2025 Gen Z protests in Kenya demonstrated the dual-edged use of AI technologies. While it played a crucial role in amplifying the grievances that the young people had against the government (Mbugua, 2025), it was also used to propagate social media disinformation to scales never seen before. The study focused on how the production and dissemination of AI generated disinformation on social media and mainly perpetuated by the young people, contribute to societal polarisation in Kenya.
Recent studies at the global level have significantly explored how generative AI has been used in social media disinformation campaigns, opportunities and challenges (Bontcheva et al., 2024; Bontridder & Poullet, 2021; Haq, 2025; Jaidka et al., 2025; Łabuz & Nehring, 2024; Lucas et al., 2024). This study still explored this area but from a unique context and study population lens. Many studies on disinformation focusing on Kenya are yet to incorporate and focus on the generative AI phenomena that was exploited by the young people popularly known as the Gen Z, during the anti-finance bill protests in the country. Authors mainly focus on how digital technologies including social media especially X have been used to propel digital activism with Ogungbade et al. (2024), establishing X as the most popular platform among the young people in Nigeria, digital technologies mediating digital resistance (Ardebili, 2025), mobilisation on X does not translate into physical mobilisation (Mwangi, 2025), digital disruption fuelling activism (Radoli et al., 2025), and from tweets to the streets (Twinomurinzi, 2024). This study investigated how generative AI was used to amplify the Gen Z protests and if this contributed to increased disinformation on social media and ultimately social divide in Kenyan society.
Figure 1. How AI-generated disinformation influences societal polarisation
Methodology
The study adopted a convergent mixed-methods design integrating qualitative and quantitative designs, as they enable the production of different forms of information to yield similar results to provide a comprehensive review of how AI-generated content shaped the digital information space. The paper adopted information warfare theory and social polarisation theory as analytic lenses to explore the double-edged role of generative AI. The study combined content analysis of AI-generated content purposively selected on X and TikTok, 13 key informant interviews for depth, and a survey for breadth, capturing public experiences, perceptions and patterns in an effort to provide a comprehensive analysis of the research problem. The mixed-method approach for this study was anchored on Creswell & Creswell (2017) observation that “the convergence of qualitative and quantitative provide additional insights beyond the information provided by either the quantitative or qualitative data alone”.
Content analysis endeavoured to identify patterns of AI-generated disinformation and thematic analysis of X posts and TikTok videos from June 2024 to August 2025. The ultimate sample size for X was 104 AI-generated posts after exhausting the relevant hashtags and meeting saturation levels where the same posts would appear across the different hashtags. For TikTok, 102 AI generated posts were sampled. The sampling frame used the popular hashtags during the protests including #RejectFinanceBill2024, #RutoMustGo, #OccupyParliament, #OccupyStatehouse, #EndAbductionsKE, #OccupyChurches, among others. #AngukaNayo, #SiriNiNumbers, #EndPoliceBrutality, #sabasaba, and #nanenane. The survey targeted 400 Gen Z participants, arrived at by using Taro Yamane’s formula, and was administered through an online Google Form shared through WhatsApp, X, Instagram, and TikTok and through face-to-face mechanisms.
Findings
The study set out to examine how generative AI applications during the Gen Z protest impact societal polarization. Findings from the survey were integrated with qualitative insights from the key informant interviews and patterns from X and TikTok data, which were then triangulated to provide a comprehensive explanation and understanding of the phenomena under study. A total of 412 responses from the survey, 13 key informant interviews, 102 TikTok posts, and 104 X posts assessed to contain generative AI features were analysed.
A conceptual flow was adopted to answer how the use of generative AI during the Gen Z protests impacted societal polarisation. Digital-AI drivers enabled information operations which led to information disorder and narrative construction, which in turn were amplified leading to polarization and then youth mobilization thereby, establishing the intersection between the variables. Four themes namely Digital-AI Structural Drivers, Information Dynamics, Amplification Mechanism, and Perception Formation and Social Polarization were derived from the research objective in relation to the data collected.
Figure 2. A conceptual flow on the Impact of AI to establish the intersection of the various variables
Digital and AI Structural Enablers
There was consensus among the key informant respondents that the digital information ecosystem of the Gen Z protest was characterised by AI-assisted content generation, decentralised communication networks with platform ubiquity, particularly X and TikTok enabling rapid circulation and wide geographic reach of content. Findings from the survey established that the sampled population mainly young people (Figure 3) consumed and met their own information needs from TikTok at 36%, from more than two platforms at 27%, X at 21%, Facebook at 9% and Instagram at 6% as illustrated in figure 4.
Figure 3. Age Distribution of the Survey Participants
The survey findings support analysis from the qualitative data where Key Informant Interviewee 01 (KII-01) observed that “people would pick that information, circulate it on online platforms; of course TikTok was very heavy on that and X, and also WhatsApp groups adding that social media is the biggest mobilisation platform in our generation and it is available for everyone”. A majority of the participants identified X and TikTok as the dominant social media platforms that played an integral role in the information flow and shaped how the protests unfolded and were perceived.
Figure 4. On which online platform did you engage with the content?
The findings were corroborated by the relatively high engagement metrics (Figure 5) of the selected TikTok and X posts deemed to be AI-generated, suggesting that the digital information space aligned with information warfare perspectives emphasising control of the information environment.
Figure 5. Average Engagement Metrics of each of the sampled posts on TikTok and X
Key Informant Interviewee 07 (KII-07) argued that “the Gen Z movement revolutionized the media industry, making the flow of information instant. It is live. No one is editing parts”. Additionally, the participant observed that “the young people would say, but why do we need to go on TV? We have smartphones. This smartphone has a screen. This screen can go online. And the guys watching TV who have social media accounts, let’s stream this thing”. This demonstrates the shift in media consumption from a centralized professional journalism practice to a decentralized, user-generated content ecosystem. Digital and AI-generated tools were weaponized by both sides of the conflict, leaving disorder and narrative contestation in the information environment (Jaidka et al., 2025; Maundu, 2025).
Information Disorder and Narrative Construction
The Gen Z protest information ecosystem reflected the character of the movement itself: organic, fast-moving, and participatory, demonstrating how modern civic action is shaped by digital connectivity. However, that information space also ushered in severe information disorder and narrative contestation. Key Informant Interviewee 08 (KII-08) captures this environment aptly by acknowledging that “the flow of information during the protests was rapid, decentralized, and largely youth-driven. Gen Z relied heavily on digital platforms to organize, share real-time updates, document events on the ground, and counter misinformation. Information moved horizontally rather than through traditional hierarchies, which allowed the movement to remain agile and difficult to suppress”. With generative AI employed in generating and producing hyper-realistic content in seconds, the prevailing information disorder and narrative contestation was quite volatile and explosive. Disinformation and misinformation defined this space, making it an information warfare conflict.
The viral hashtags during the Gen Z protests, starting with #Rejectfinancebill, #Rutomustgo, #Endabductions, #Occupyparliament, #zakayoshuka, #occupystatehouse, among others, give a glimpse into the narrative contestation at the time. Key Informant Interviewee 02 (KII-02) acknowledged that Kenya is a fast-developing country and it is therefore “vulnerable to a lot of conspiracies, misconceptions because of the internet. So, our young people are very quick to consume what they want to consume online and what they are given to consume”. This demonstrates how well resonant and emotional narratives spiced with some disinformation or misinformation are quick to spread. Findings from analysis of TikTok and X AI-generated content revealed that dominant narratives evolved around perceived poor leadership, injustices, call to action, distrust, greed and corruption, and policy criticism, as shown in Figure 6. However, the percentages between the two platforms are inconclusive, as platforms, especially TikTok, have been taking measures to remove content violating its community guidelines, thereby limiting the volume of content sampled (Oyiela, 2026).
Figure 6. Dominant themes from AI-generated content on X and TikTok
Information Disorder and Narrative Construction manifested in four key ways, including through the production of AI-generated content creation (deepfakes, synthetic media, captions), disinformation and misinformation, selective framing of content, and crafting emotionally charged narratives.
Perception Formation and Social Polarisation
The information disorder and contested narratives environment during the protests considerably, shaped perceptions and exposed existing tensions across multiple axes, leading to a highly charged and polarised Kenyan society. Disinformation and AI-generated content intensified polarisation during the protests by amplifying divisions across generational, political, institutional, and even ethnic lines.
The information space during the protests was defined by insults, body shaming, provocative AI-generated content, misinformation, and in-group and outgroup identities, leading to a very divided people, supporting the assumptions of social polarisation theory. Revelations from a majority of the key interview informants revealed that the protests saw heightened fragmentation and polarisation between different groups such as the citizens against the police, the Gen Z against other generations, and ethnic divide. Findings from the survey were quite mixed on whether AI-generated content contributed to division or polarisation during the protests with 41% of the respondents in agreement while 34.5% disagreed and a further 24.5% being unsure, as presented in Figure 7.
Figure 7. Perception on whether AI-generated content contributes to division or polarization
This is explained by findings from the survey where 65.3% (Figure 10) of survey respondents indicated their ability to identify AI-generated content and would probably question the authenticity of such information. Perception formation and social polarisation were displayed during the Maandamano (protests) through lack of trust and distrust in institutions, “Us vs them” identity formation, and establishment of echo chambers. The Gen Z protests collectively brought to the fore the perception that the government was not listening to its people, the government here representing all state institutions from the Presidency down to the local police post in far-flung villages. The majority of the key informants and survey responses unanimously agreed that public trust in government institutions such as the police, parliament and the presidency is very low.
Key Informant Interviewee 12 (KII-12) observed that “the images of police handling the Gen Z, some of it, and even the death of some of these young people, I think it left a very sour taste on many people”. Key Informant Interviewee 03 (KII-03) added that “people have continued to lose trust with government institutions even more. The parliament, the presidency, all manner of leaders, people just don’t trust them”. This is partly because Gen Z had the narrative, as noted by KII-02 that “we are peaceful, we are patriotic, they come with the flag; if there is a trigger then it is the police, not us; we are peaceful protestors”. The protests produced strong emotional orientation towards state actors, especially against the police, demonstrating the weaponisation of the information environment, which is a key component of information warfare theory.
For an overwhelming majority of the participants, the depiction of the president lying dead in a coffin reflected the magnitude of the loss of trust Kenyans have in their President. KII-12 noted that “it was a symbolic loss of trust, a loss of legitimacy of the government”. The image and its numerous variations drew polarizing reactions across the survey, the interviews, and, as evidenced through the high engagement metrics, the image and its variations generated on both X and TikTok. Reactions to the image were as polarizing as the image itself. For some, like Key Informant Interviewee 03 (KII-03) “the image marked the height of defiance”, the courage, and the willingness of Gen Z to push boundaries. For others, “it was very bad, shocking”, and it had a negative impact on the overall support of the movement. This led the movement to become about “Us versus the government. Us versus someone else” demonstrating the heightened polarisation.
Role of AI in Amplification and Manipulation
Generative AI was instrumental in the production and creation of content, including distorting and manipulating it, partly leading to the polarised information space during the protests. Findings from the survey projected that about 52.2% (27.43% + 24.76%) of the survey respondents agreed that AI-generated content had some significant influence in shaping public perception negatively as shown in Figure 8. This may be attributed to the high volume of AI-generated content shared on X and TikTok, with polarising hyper-realistic images and videos of political leaders placed in coffins, undressed, or made to say statements they have never said and do things they have never done, thereby creating a false impression of reality.
Figure 8. Shows the level of agreeableness if AI-Generated content increased public confusion during the protests
As Key Informant Interviewee 07 (KII-07) remarked, “AI made us believe that we could operate outside the boundaries of modernity and it was acceptable”. A significant volume of the AI-generated content, especially videos posted on TikTok, was comical and utilised dark humour to encourage protest participation, call out poor governance and leadership, highlight injustices against the youths, illustrate their discomfort with the government, castigate greed and corruption, and pork holes in government policies like the finance bill and issues bedevilling the health sector, as earlier illustrated in Figure 5.
AI-generated content was a mechanism for political satire, construction of protest narratives, undermining political legitimacy, and a driver of emotionally charged digital activism. Key Informant 10 (KI10) observed that AI helped Gen Z get information they did not even know about the government, attributed to the ease of using the prompt function of AI. A Likert scale where respondents selected their perception on whether AI-generated disinformation impacted public opinion found that 73% (40%+33%) were in agreement, revealing how powerful AI-generated content is in shaping issues, as shown in Figure 9.
Figure 9. Perceived impact of AI-generated disinformation on public opinion
However, Key Informant Interviewee 04 (KII-04) averred that “generative AI was prominently being used to share ethnic content, hate speeches, misinformation. Basically, disinformation tries to share content which was basically fake and not true”, thereby contributing to the volatile information environment.
AI-generated content sampled on TikTok and X incorporated emotional triggers, symbolic imagery, national identity framing, protest legitimacy, crisis amplification, and us versus them narratives to influence the masses. The AI content produced was packaged in different ways, including deepfakes, synthetic audio, images and videos, memes, symbolic visuals, and hybrid content. KII-12 substantiated that “there was a lot of the use of generative images and deepfakes and even synthetic audio during this particular protest. We saw this in synthetic audio songs calling for the youths to come out and uh reject the finance bill and send the president out of State House. We also saw the use of AI-generated images of leaders, which were pejoratively created to erode the legitimacy of the government and its leaders”.
Findings from the survey (Figure 10) showed that AI-generated content after all may not be so hard to detect, with 65.3% of the respondents indicating their ability to distinguish between AI-generated and authentic content, while 11.9% were unable, with a further 22.8% uncertain about their ability to differentiate between AI-generated and authentic content. While the relatively high percentage of those able to recognise AI-generated content denotes a high level of awareness and digital literacy, the combined percentage of 34.7% of those unable and uncertain to pick out AI-generated content points to the possibility that a sizeable number of the population is vulnerable to manipulated synthetic media.
Figure 10. Respondents’ ability to differentiate between AI- generated and authentic Content
The deepfake image of President William Ruto was easily identified by respondents from both the survey, key informant interviews and from assessment of the X and TikTok posts as AI-generated. Other deepfakes, as described by Key Informant Interviewee 11 (KII-11), were “seeing people using AI to put male politicians in women’s clothes, bikini, trying to degrade them”. KI4 claimed that “everyone was trying to distort the images and generate images where you have naked politicians or politicians in coffins or being hanged or being crucified by the citizens”. This explains the proliferation of deepfakes that targeted mainly the political class, heightening the already heavily polarised environment, pointing to information warfare tactics exploited at the time.
Conclusion and Study Implications
Findings from the survey projected that 52.2% of the survey respondents, with a majority being young people at 79.4%, agreed that AI-generated content had a significant influence in shaping public perception negatively. Findings also revealed that information exchanges happened mainly through TikTok at 36%, and X at 21%, while 27% used more than two social media platforms to consume AI-generated messages. These revelations show the close intersection between generative AI, youth-led movements, social media disinformation, and societal polarisation. Furthermore, the findings established that disinformation and AI-generated content intensified polarization during the protests by amplifying divisions across generational, political, institutional, and even ethnic lines. This led to an “us versus them divide” not only within the different fragmentations in the movement, but also across the wider public, illustrating the dynamics of social polarisation and affective group identity formation. Studies show that social media and other emerging technologies, including AI, increase information asymmetries, which erode shared truths, strengthen identity boundaries, and reinforce the formation of echo chambers, thereby intensifying prevailing social and political divides in societies (Castells, 2011).
References
Ardebili, S. (2025). Tweets of Resistance: Social Media and Mobilization in Contemporary Kenya A discursive study on the Kenyan protests and online activism in 2024. https://lup.lub.lu.se/student-papers/search/publication/9201474
Bontcheva, K., Papadopoulous, S., Tsalakanidou, F., Gallotti, R., Dutkiewicz, L., Krack, N., Teyssou, D., Severio Nucci, F., Spangenberg, J., & Srba, I. (2024). Generative AI and disinformation: Recent advances, challenges, and opportunities. https://lirias.kuleuven.be/retrieve/758830
Bontridder, N., & Poullet, Y. (2021). The role of artificial intelligence in disinformation. Data & Policy, 3, e32.
Castells, M. (2011). The power of identity (Vol. 14). John Wiley & Sons. https://books.google.com/books?hl=en&lr=&id=l9asXV40MegC&oi=fnd&pg=PP1&dq=Manuel+Castells%E2%80%99+Identity-Based+Polarisation+in+the+Network+Society&ots=nhl5fqez1K&sig=DMpfz4F7rx-myg-A2mjYHsBKfRk
Haq, A. (2025). The Weaponization of Generative AI in Disinformation Campaigns. Global Political Review, X(III), 165–174. https://doi.org/10.31703/gpr.2025(X-III).15
Jaidka, K., Chen, T., Chesterman, S., Hsu, W., Kan, M.-Y., Kankanhalli, M., Lee, M. L., Seres, G., Sim, T., Taeihagh, A., Tung, A., Xiao, X., & Yue, A. (2025). Misinformation, Disinformation, and Generative AI: Implications for Perception and Policy. Digital Government: Research and Practice, 6(1), 1–15. https://doi.org/10.1145/3689372
Kariuki, T. N., & Gichanga, R. (2025). Social Media Coordination as a Strategic Messaging Platform for Political Advocacy in Kenya: A Case Study of Kenya’s Gen Z uprising in 2024. African Multidisciplinary Journal of Research, 2(2), 100–124.https://doi.org/10.71064/spu.amjr.2.2.2025.430
Łabuz, M., & Nehring, C. (2024). Information apocalypse or overblown fears—What AI mis‐ and disinformation is all about? Shifting away from technology toward human reactions. Politics & Policy, 52(4), 874–891. https://doi.org/10.1111/polp.12617
Lucas, J. S., Maung, B. M., Tabar, M., McBride, K., & Lee, D. (2024). The longtail impact of generative ai on disinformation: Harmonizing dichotomous perspectives. IEEE Intelligent Systems, 39(5), 12–19.
Maundu, C. (2025, July 12). How Kenyans are using AI during protests. https://globalvoices.org/2025/07/12/how-kenyans-are-using-ai-during-protests/
Mbugua, J. (2025). Why Kenya’s Gen Z Has Taken to the Streets. https://www.journalofdemocracy.org/online-exclusive/why-kenyas-gen-z-has-taken-to-the-streets/
Mwangi, E. N. (2025). The Twitter/X Revolution: Gen Z’s Political Participation in Kenya’s Digital Age [Master’s Thesis, Temple University]. https://search.proquest.com/openview/1d1ac7ee3182784ede634e9d1a01fae1/1?pq-origsite=gscholar&cbl=18750&diss=y
Oyiela, J. (2026, February 17). TikTok deletes over 580,000 videos in Kenya amid stricter community guidelines. https://www.tnx.africa/tech/article/2001541084/tiktok-deletes-over-580000-videos-in-kenya-amid-stricter-community-guidelines
Radoli, L. O., Eboi, A. A., Aswani, D. R., & Nakhumbi, R. O. (2025). X Space and the Revolution of Digital News Content: The Case of Generation-Z Protest Narratives in Kenya. African Multidisciplinary Journal of Research, 1(1), 602–621.
Twinomurinzi, H. (2024). From Tweets to Streets: How Kenya’s Generation Z (Gen Z) is Redefining Political and Digital Activism. African Conference on Information Systems and Technology. https://digitalcommons.kennesaw.edu/acist/2024/presentations/13
OpenEdition vous propose de citer ce billet de la manière suivante :
Hillary Bobo (22 juillet 2026). Generative AI, Social Media Disinformation, and Polarization During the 2024 Gen Z Protests in Kenya. Mambo ! Consulté le 8 octobre 2026 à l’adresse https://doi.org/10.58079/16lw9
