Author: Berend Watchus. Independent non profit AI & Cyber Sec Researcher. March 14, 2026. [Publication for OSINT TEAM]
LROO Rug Pull Detector: A Leakage-Resistant Framework Based on On-Chain and OSINT Signals
Researchers at Ilam University and VAISR Research Group recently published a rug pull detection framework that quietly confirms something OSINT practitioners have long suspected: the scam is visible before the money disappears.
Their system combines blockchain data with just two open-source signals — tweet volume and Google search spikes — and achieves 98.2% detection accuracy. More importantly, these signals appear days or weeks before any on-chain collapse occurs. The liquidity drain is the last act. The internet already told the story.
This matters because it validates a core OSINT principle: behavioral patterns precede events.
Coordinated promotional campaigns, artificial sentiment engineering, manufactured search interest — these leave traces. They are operational signatures, not noise.
The second finding is equally striking. Two signals. That’s the entire OSINT layer in a published, peer-reviewed detection framework achieving near-perfect detection. No Telegram growth velocity, no Discord monitoring, no coordinated posting pattern analysis, no influencer mapping.
And it still hits 98.2%.
That’s not an argument for doing less. It’s an argument for how much signal is already sitting in plain sight — uncollected, unmodeled, ignored. If two crude proxies for human attention already catch scams before the blockchain does, the internet is a better fraud detector than the chain itself.
OSINT isn’t support material here. It’s the lead.
OSINT SIGNALS: Before the Rug Pulls, the Internet Already Knew was originally published in OSINT Team on Medium, where people are continuing the conversation by highlighting and responding to this story.