Graph AI Tracks Crypto Crime Across Networks

VisionX AI's AEGIS model is a commercial launch whose detection performance has not yet been independently benchmarked.

Updated on Nov. 20, 2026 in Artificial Intelligence

Graph AI Tracks Crypto Crime Across Networks

VisionX AI, a Singapore-based AI-native RegTech firm operating within Alpha Ladder Group, has launched AEGIS, a proprietary Large Graph Model for detecting and investigating financial crime, along with TraceX, the first product built on it for blockchain investigations. The launch is a commercial product release, not a settled industry standard.

Why it matters

Illicit finance is increasingly structured across networks of virtual assets, which the Financial Action Task Force cautioned in July 2026 is being used for fraud, sanctions evasion and money laundering. AEGIS reflects a move from assessing individual payments toward modelling the transaction graph as a whole, a shift in how compliance tooling is being built.

AEGIS is built on a Graph Transformer architecture and pretrained on transaction graph data, while VisionX KYT reports coverage of 1 billion wallets on 25 chains. The source provides no independent accuracy or performance benchmarks, and the coverage figures are company-reported.

The players

VisionX AI

A Singapore-based AI-native RegTech firm that builds graph-based financial crime detection, operating as an independent business within Alpha Ladder Group and incubated inside MetaComp.

Financial Action Task Force

The intergovernmental body that sets international standards for combating money laundering and terrorist financing, which warned in July 2026 about criminal use of virtual assets.

The details

AEGIS treats financial activity as a graph, where wallets and transactions are nodes and edges, so it can spot suspicious patterns across a network rather than judging each payment alone. VisionX pairs this graph model with large language models from established vendors in a hybrid set-up: the graph model supplies network intelligence, and the language models handle natural-language queries and reasoning. The system is pretrained on transaction graph data, which is the material it learns network patterns from. The product targets illicit flows that move through stablecoins, cross-chain bridges, decentralised exchanges and mixers.

Timeline

  1. November 2025: StableX Network debuted.

  2. July 2026: The Financial Action Task Force cautioned that criminal groups use virtual assets for fraud, sanctions evasion and money laundering.

  3. November 20, 2026: VisionX AI launched AEGIS and TraceX.

The Tech Race

The launch positions graph-based transaction modelling against the conventional approach of rules-based screening of individual payments. The source does not name competing graph-based compliance products, so the competitive comparison is not established.

Compliance teams at financial institutions and virtual asset firms are the first likely users, since the product targets fraud, scam prevention and investigations. The source gives no pricing, availability dates or access terms for AEGIS or TraceX.

The takeaway

Watch for independent evaluations of AEGIS detection accuracy, which the launch announcement does not include. The FATF's next publications on virtual asset risks will indicate whether regulators treat network-level analytics as a recognised compliance approach.

Further reading

For more on how machine learning is being applied to detection problems, explore our Artificial Intelligence coverage.