| Source | Period | Volume | Notes | |--------|--------|--------|-------| | Public Telegram groups (e.g., Kang‑Crew, Prank‑Ops) | Jan 2023 – Jun 2024 | 3,212 messages | Scraped via Telethon API (rate‑limited). | | Discord servers (e.g., Prank‑Lab, Meme‑Factory) | Jan 2023 – Jun 2024 | 4,567 messages | Bot‑generated logs, anonymised. | | Victim surveys (n = 102) | Mar 2024 – May 2024 | N/A | Distributed via Reddit’s r/Assistance. | | Semi‑structured interviews (n = 27) | Apr 2024 – Jun 2024 | N/A | 12 victims, 9 agents, 6 platform moderators. |

| Actor | Primary Motivations | |-------|---------------------| | Customers | Entertainment, status‑seeking, “revenge” against perceived slights. | | Agents | Monetary gain (average $12 USD per prank), community belonging, skill development (deep‑fake tech). | | Victims | Mostly involuntary; some “participated” after being duped into a “challenge”. |

| Idea | How It Works | |------|--------------| | Premium Crown Skins | Sell exclusive avatar outfits (e.g., Golden Dragon Crown). | | Indo‑18 Content Packs | Monthly subscription for fresh local meme packs, exclusive AR filters, and priority entry to “Indo‑18” tournaments. | | Branded Prank Templates | Partner with brands for “sponsored pranks” (e.g., a soda company’s “Fizz‑Burst” sound). | | Data‑Driven Insights | Offer anonymized trend reports to marketers on what prank categories are trending per region. |

If you suspect that a service is violating any of these statutes (e.g., requesting money under false pretences, threatening to expose private content, or distributing illegal porn), you should report it to the Indonesian National Police’s Cyber Crime Unit (Polri) or the Ministry of Communication and Information Technology.

Prank‑Kang illustrates how commodified harassment can be packaged as a service, complete with pricing tiers, SLA‑like guarantees (“prank delivered within 24 h”), and reputation systems akin to gig‑economy platforms. This challenges traditional dichotomies of harassment vs. humor by foregrounding transactional motives.

Current moderation pipelines, built on keyword‑based detection, fail to capture context‑rich, scripted pranks that masquerade as benign memes. The cross‑platform nature of Prank‑Kang (Telegram → Discord → TikTok) also exposes a siloed enforcement problem.

| Layer | Description | Tools / Technologies | |-------|-------------|----------------------| | Target Acquisition | Automated scraping of public profiles (TikTok, Instagram) using keyword filters (“celebrity”, “contest”). | Python + Selenium, OpenCV for facial detection. | | Scripted Execution | Pre‑written “prank scripts” (e.g., fake DM from a brand, deep‑fake voice call). | FFmpeg for video morphing, Resemble.AI for voice synthesis. | | Delivery Channel | Multi‑modal (link drops, Discord voice pings, Telegram bots). | Custom Telegram bot (“@prank_kang_bot”). | | Reputation Recycling | Post‑prank “proof” (screenshots, reaction videos) posted to MiraIndira Indo18 channel, earning “karma points”. | Discord webhook integration, auto‑posting to TikTok via API. |

Core agents (top 5% by betweenness centrality) handle payment processing (via crypto wallets) and script customization.