Key points are not available for this paper at this time.
Stablecoins carried most illicit virtual-asset transaction value in 2025, while many blockchain anti-money laundering (AML) tools still reflect design assumptions inherited from Bitcoin’s unspent transaction output graph. Evidence from peer-reviewed studies, preprints, industry reports, and regulatory sources published from 2019 to early 2026 was reviewed to assess how far those assumptions carry into USDT and USDC environments. Commercial platforms support attribution, wallet-risk scoring, exchange screening, and investigations, but cross-chain routing weakens continuous tracing; Elliptic estimated 21. 8 billion in illicit and high-risk flows through cross-chain methods in 2025. Bitcoin classifiers report accuracies above 97% on Elliptic-family benchmarks, although class imbalance and limited precision-at-recall reporting make operational value difficult to judge. StableAML, the only USDT/USDC-specific classifier identified, found stronger signal in behavioral wallet features than in topology-heavy Bitcoin-derived approaches. Regulation and issuer freezes add intervention points, but slow response cycles, scarce stablecoin labels, and closed vendor thresholds limit independent evaluation. Stablecoin AML evaluation depends on public labeled datasets, precision-at-recall metrics, latency measures, and independent audits of commercial tools.
Hamed et al. (Wed,) studied this question.