Reproduce “Victim as a Service: Designing a System for Engaging with Interactive Scammers” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Daniel Spokoyny, Nikolai Vogler, Xin Gao, Tianyi Zheng, Yufei Weng, Jonghyun Park, Jiajun Jiao, Geoffrey M. Voelker and 2 more

CHATTERBOX is a system for collecting conversations with unsolicited text-based scammers whose criminal purpose is not obvious at first contact. It uses accounts to attract scam attempts, then an LLM-driven persona, with human oversight, to sustain engagement and migrate conversations to WhatsApp. In a seven-week deployment, it engaged 568 scammer accounts; this demonstrates feasibility, not scam prevention or changed outcomes.

Key insights

  • Automated, long-term engagement with interactive scammers using LLMs is feasible and scalable, providing a previously unavailable source of empirical data on complex fraud lifecycles.
  • Interactive scams typically involve prolonged trust-building phases, multimodal communication, and early migration to encrypted messaging platforms before monetization is attempted.
  • Scammers actively probe victim authenticity, personalize interactions, and primarily employ liking, affinity, common ground claims, and appeals to benevolence as trust-building tactics.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.