Steam Bots Unmasked: Behavioral Profiling and Economic Incentives of Automated Accounts

Abstract

The rise of automated accounts on the Steam platform undermines ecosystem integrity and fuels a hidden economy of fraud. This study provides a comprehensive examination of the Steam bot ecosystem through two datasets: 41,472 public Steam profiles from a bot-prevalent community and 1,103 account listings from a secondary resale marketplace. Unsupervised K-Means clustering across 12 behavioral features achieved robust separation between automated (n=38,707) and legitimate (n=2,765) accounts (silhouette score 0.57). BOT accounts exhibited near-zero levels, default visuals, templated names, and concentrated playtime in Counter-Strike 2 and Banana. Beyond behavioral profiling, transaction analysis revealed a three-tier money laundering ecosystem: (1) external carding, converting stolen payment cards to Steam Wallet Codes via grey-market platforms lacking 3D Secure; (2) internal consolidation, draining the account balance through overpriced market purchases from operator-controlled accounts; and (3) gameplay-based extraction, where bots farm tradeable items and transfer them to centralized master accounts for net profit. The dominance of HKD-denominated Steam Wallet Codes further evidences coordinated, carding-funded provisioning. These findings provide a robust behavioral and financial fingerprint of automated accounts, offering actionable insights for platform governance and fraud detection in digital ecosystems.