Is TikTok Pushing Kids Toward WAR?!

Photo: XanderSt / Shutterstock

When armed groups modernize their playbook, the harm scales with the network they plug into; in Colombia, the recruitment of children now travels the same algorithmic rails that deliver dance trends and family photos—making platform governance a frontline child-protection issue, not a PR sidebar.

At a Glance

Human Rights Watch reports that Colombian armed groups use Facebook and TikTok to reach, groom, and recruit minors, via public posts and direct outreach.
Colombia’s Special Jurisdiction for Peace attributed 44% of child recruitment in 2025 to social media channels, signaling a structural, not anecdotal, shift.
Meta and TikTok say their rules prohibit such content and removed accounts flagged to them; HRW’s core finding is about what still circulated before takedown.
The social feed is only the final mile in a wider recruitment pipeline fueled by local insecurity, poverty, and weak state presence—conditions armed groups exploit.

What HRW’s evidence shows: social media as a recruitment multiplier

Human Rights Watch’s 102-page investigation details how Colombian armed groups operationalize mainstream platforms for child recruitment: slick propaganda that normalizes armed life; deceptive job offers that move conversations into direct messages; and contact funnels that connect minors to local operatives. The report’s linchpin statistic is not HRW’s own—Colombia’s Special Jurisdiction for Peace assessed that 44 percent of minors recruited in 2025 were recruited through social media. The mechanism is depressingly familiar to anyone who studies digital mobilization: attention capture at scale plus low-friction outreach equals a surge in “top-of-funnel” prospects, many of them children with few alternatives and a phone in their hand.

That proportion should be read for what it is: a signal that recruitment has adapted to the attention markets where Colombian youth actually live. Platforms that optimize for engagement can—absent aggressive guardrails—rank and recommend content that glamorizes group identity, risk-taking, or easy money. The HRW material catalogues the pattern, from videos that aestheticize armed life to accounts that advertise “security work” or “quick cash,” then slide into private messaging to close the deal. This is not a lone bad post slipping through moderation; it is the use of industrial-scale distribution to find vulnerable children at a cost per contact that analog-era recruiters could only dream of.

Why the trend accelerated: conflict dynamics met algorithmic distribution

The digital story cannot be disentangled from Colombia’s security landscape. Since the 2016 peace agreement, the country has seen fragmented armed actors compete for territory and revenue; the demand for recruits rose alongside localized violence. Multiple monitoring efforts—UN-verified data, humanitarian assessments, and conflict analysis—converge on a stark picture: child recruitment has climbed sharply over the past five years as groups mix old coercive methods with new digital lures. Social platforms did not invent the demand; they reduced the search and persuasion costs and weakened geographic constraints, allowing recruiters to cast a wider net into precisely the communities where the state is least present and prospects most precarious.

From a practitioner’s lens, this is classic funnel engineering. Public-facing content performs identity work and normalizes contact; short-form video lowers cognitive load and raises emotional salience. The call to action moves to DMs, WhatsApp, or encrypted apps where screening, pressure, and logistics proceed. Each stage exploits platform affordances originally designed for commerce and community. When nearly half of documented recruitment is routed through social media, the causal leverage of distribution and discovery systems is no longer theoretical—it is operational.

What the platforms say—and what that actually addresses

Meta and TikTok point to policies that prohibit content promoting violent organizations and to moderation systems—automated and human—that detect and remove violations. Both companies removed the profiles HRW flagged, and TikTok says it permanently banned them. These responses matter; rules set the baseline, and takedowns narrow exposure windows. But the heart of HRW’s claim is not that no rules exist; it is that harmful content reached minors at scale before enforcement triggered, and that recommendation engines appeared to amplify some of it. Removals after external notification confirm policy; they don’t rebut the documented exposure or the recruitment that begins in the time-to-takedown gap.

TikTok adds that it has not observed a platform-specific pattern of groups directly targeting minors and frames the issue as cross-platform. Both points can be true without diminishing the risk. Recruitment pipelines are indeed cross-platform—public content on one service, contact moved to another—but that fact distributes responsibility; it does not dissolve it. For high-harm categories involving minors and violent actors, the operative standard is not eventual removal; it is prevention of reach, demotion of borderline content that functions as a lure, and rapid suppression of networks that repeatedly reconstitute under new handles.

The mechanics to fix: discovery, provenance, and speed

When risk involves minors and violent organizations, three technical levers matter most. First, discovery suppression: deprioritize or block recommendation pathways to accounts and content that match armed-group aesthetics, narratives, or known recruitment tropes, even when individual posts are crafted to skirt explicit policy terms. This is a ranking problem, not merely a rules problem. Second, provenance mapping: continuously map clusters of accounts, audio tracks, hashtags, and contact numbers that co-move with identified recruitment operations; kill the cluster, not just the post. Third, latency: collapse time-to-takedown from hours to minutes for this category using high-recall classifiers triaged by specialized safety teams with Spanish-language, Colombian regional, and conflict-dynamics expertise. These are routine in child sexual exploitation countermeasures; they should be table stakes here as well.

None of this negates the need for due process and error correction. But the cost of false negatives—children funneled into armed groups—justifies conservative, precautionary tuning of ranking and discovery in affected regions. The HRW record, paired with Colombian judicial and UN-linked data, establishes the harm profile clearly enough to warrant that shift.

Beyond platforms: the upstream pressures that make lures convert

Digital exposure is a necessary condition for a large share of today’s recruitment; it is not a sufficient one. Children join because promises—money, status, protection—land in contexts where schools underperform, legitimate work is scarce, and armed actors dominate local life. Evidence from humanitarian and conflict-analysis organizations shows recruitment risk concentrates where state services are thin and violence high. Effective mitigation therefore means coupling platform reforms with offline interventions: cash transfers or scholarship programs targeted to at-risk municipalities, rapid-response child-protection teams that intervene when online grooming begins, and local outreach that competes on the same narratives—belonging, pride, income—without the coercion.

For governments and donors, the policy implication is straightforward. Treat social-media recruitment as a force multiplier that can be dialed down through product changes, but don’t mistake it for the engine. Fund the boring, durable upstream work—schools, safety, family support—while pressing platforms for verifiable transparency on detection rates, recommendation demotion, and time-to-takedown in Colombia’s conflict-affected corridors. The combination, not either domain alone, bends the curve.

Human Rights Watch says Facebook and TikTok have struggled to prevent content linked to Colombian armed groups from reaching children, including posts that promote recruitment and portray armed-group life positively.#HumanRightsWatch #Colombia #ChildSafety #SocialMedia… pic.twitter.com/cyi6xLokSP

— Pakistan TV (@PakTVGlobal) August 24, 2026

Accountability that matches the stakes

HRW’s findings are not a culture-war critique of “big tech”; they are a child-protection brief built on Colombian institutions’ own figures and a decade of conflict observation. The counterpoints from Meta and TikTok acknowledge policy intent and post-hoc enforcement; they do not erase the pre-enforcement reach that recruitment content achieved. The standard, then, should be measurable outcomes: fewer minors reached, fewer leads converted, faster removal, and demonstrable deranking of recruitment-adjacent content. Where nearly half of recruitment flows through social media, anything less than outcome-level accountability leaves the core risk untouched.