![\"bbr](\"https:\/\/torrentfreak.com\/images\/bbr.png.webp)
<\/picture><\/center> <\/p>\nThe findings reveal that piracy is certainly not a fringe activity on Telegram. On the contrary, it is massively popular.<\/p>\n
4.85 billion post views for 19,033 titles<\/h2>\n
To map the ecosystem, the researchers relied on a locally run large language model to label posts. This helped them to identify 19,033 unique pirated titles across various Telegram channels, including 14,632 movies and 4,401 TV shows produced by 3,941 companies.<\/p>\n
As on regular pirate sites, anime is rather popular. The most pirated rightsholder is Japan\u2019s Toei Company, home to One Piece and Dragon Ball, which accounted for 17% of the titles. As shown below, Netflix is in second place with 15%, followed by Warner Bros. at 12.4%.<\/p>\n
<\/p>\n
![\"top](\"https:\/\/torrentfreak.com\/images\/toppirated1.png.webp)
<\/picture><\/center> <\/p>\nThese numbers get more context when looking at the total views. According to the researchers, the \u2018pirate\u2019 posts were found on 983 channels where they amassed 4.85 billion views.<\/p>\n
The views are not per title, as a single post can include more titles. Nonetheless, the researchers estimate a total loss of $17.49 billion, with United States content accounting for $8.17 billion and Japanese content $3.72 billion.<\/p>\n
This is a loose estimate, assuming that 1% of the views translate into lost sales, based on the cheapest legal option available. Also, the researchers capped lost sales at a single subscription cost when multiple titles from one service were linked.<\/p>\n
Built to Survive Takedowns<\/h2>\n
One of the most noteworthy findings is that piracy channels use a wide variety of distribution techniques, with content scattered across interconnected channels, bots, and backup accounts.<\/p>\n
Roughly 94% of the AI-mapped channels were connected to at least one other and many of these were unfindable using traditional searches.<\/p>\n
\u201cWe also find that this ecosystem is deliberately engineered to be resilient against takedown efforts, frequently redirecting users through chains of intermediary channels and automated bots that collectively handle hosting, access control, monetization, and channel discovery.\u201d<\/p>\n
<\/p>\n
![\"telegram](\"https:\/\/torrentfreak.com\/images\/teleexample.png.webp)
<\/picture><\/center> <\/p>\nMost pirate links pointed to external hosting platforms such as TeraBox, Terashare, and GoFile. Torrents and magnet links, meanwhile, were a rarity, and the researchers only spotted nine of these links in their research.<\/p>\n
In addition to posting links to pirated content, some channels also shared compromised Netflix, Hulu, Disney+, and Crunchyroll logins, and VPN tutorials to help people bypass blocking measures.<\/p>\n
Anti-RIP: AI Powered Channel Hunting<\/h2>\n
The researchers went beyond simply mapping the ecosystem. Their findings also motivated the development of \u201cAnti-RIP,\u201d a real-time AI-powered tool that can detect video piracy on Telegram.<\/p>\n
To catch channels before they grow, the researchers generated candidate Telegram handles and probed them to see which ones were linked to piracy communities. Between February 3 and April 10, 2026, the tool scanned 249,133 newly discovered channels.<\/p>\n
<\/p>\n
![\"anti-RIP\"](\"https:\/\/torrentfreak.com\/images\/antiripframework1.png.webp)
<\/picture><\/center> <\/p>\nFrom that sweep, Anti-RIP flagged 802 piracy channels with a median age of less than 5 days, along with 299 connected channels and 108 bots.<\/p>\n
Rather than sending bare links, the team compiled the findings into evidence reports that paired each flagged channel with contextual labels describing what it was doing, from hosting and redirecting to monetizing content. These reports were sent to Telegram\u2019s abuse department as well as 17 major U.S. rightsholders.<\/p>\n
The research notes that 14 of the 17 US studios acknowledged the reports, and 4 explicitly stated that the contextual labels helped them assess and prioritize the notices.<\/p>\n
\u201cOver a 61-day period, the framework facilitated the takedown of 524 previously unknown piracy channels and 71 bots,\u201d the paper reads. Additionally, Telegram removed many flagged posts.<\/p>\n
AI Tool Isn\u2019t Flawless<\/h2>\n
Anti-RIP\u2019s reports produced measurable results. Within two weeks, 524 of the 1,101 reported channels had become inaccessible, and Telegram removed many individual flagged posts on top of that.<\/p>\n
The AI tool is far from perfect, and the researchers acknowledge that it produces false positives. When two coders reviewed a random sample of 1,000 posts used to validate the system, they found that the model had wrongly flagged 4 legitimate posts as piracy.<\/p>\n
The detection model built for the live tool is reported to be 98% accurate in testing. That figure comes from a controlled test set, however, and the paper does not publish a verified error rate for the channels that were flagged during its real-world run.<\/p>\n
The researchers have open-sourced Anti-RIP and released the dataset publicly\u00a0through GitHub<\/a>. This means that Telegram and rightsholders can put it to use, if they like. Similarly, pirates will likely use AI tools to evade detection, triggering an AI-driven game of cat-and-mouse.<\/p>\n \u2014<\/em><\/p>\n A copy of the paper, \u201cBinge, Bot, Repeat: Unpacking the Ecosystem of Video Piracy on Telegram,\u201d is available\u00a0here<\/a>. It is a preprint that hasn\u2019t been peer-reviewed yet.<\/em><\/p>\n Source: \u00a0![\"TorrentFreak\"](\"http:\/\/worldjusticenews.com\/news\/wp-content\/uploads\/2016\/11\/torrentfreak.png\")
<\/a> TorrentFreak.com<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"