Illicit Crypto Isn’t Uncontrollable: Why I’m Cautiously Optimistic

What the next phase of countering illicit finance will look like

No one has ever accused me of being an optimist — especially if they’ve read Between the Lines.

However, today I’m staking out a position of cautious optimism on one of the hardest problems in financial intelligence: getting a handle on illicit crypto flows, especially when it comes to designated terrorist groups.

To be clear: nothing in this field will ever be perfect. Threat finance is adaptive by design. But after tracking this space for years, I’m convinced the scales are tipping in ways that make overwhelming pessimism harder to sustain.

Here’s why.

Data aggregation

Every crypto transaction leaves a digital trace. Even if pseudonymous, it’s still a ledger entry — vastly different from the opaque networks of hawala or bulk cash smuggling.

The successful takedown of illicit Monero transactions by Japanese National Police is proof that anonymity tech can be pierced when regulators and enforcement pair data with persistence. Practitioners in the blockchain analysis space have told me that anonymous tokens are not as anonymous as some would have you believe.

The problem right now is that it takes a lot of people to find and process the data needed to track illicit flows and the industry struggles to keep up. That’s where AI comes in — to turn the overwhelming volume of raw data into actionable detection.

AI development

Several blockchain analytics firms are already layering machine learning into their platforms — using AI to catch anomalies humans miss. But they are not the only ones.

The Defense Advanced Research Projects Agency (DARPA) is currently developing a programme called Anticipatory & Adaptive Anti–Money Laundering (A3ML). In a recent podcast, the A3ML project lead commented his team’s goal is to make money laundering “too expensive” by dramatically increasing the effectiveness, speed, and scale of detection using artificial intelligence (AI). Because money laundering is defined by scale, speed, and complexity — exactly the domains where human investigators are outmatched and algorithmic detection excels — AI could be a genuine gamechanger.

DARPA has been quietly promoting A3ML for months, meaning it is well past the idea stage. With leading analytics firms already incorporating AI into their products, the use of AI in this field is no longer hypothetical. Given the rapid speed of advancement in current AI technologies, the efficacy of these products is likely to improve dramatically in the coming months and years.

Regulatory reform

The European Union’s MiCA regime, the U.S. GENIUS Act, and Hong Kong’s stablecoin ordinance all point to a growing global consensus that crypto markets can and must be regulated. The traction my recent articles on unregulated crypto regimes have gained underscores that this isn’t just a policy shift — there’s real appetite across the field to understand and address these risks.

I believe enforcement will continue moving toward a preponderance standard — acting whenever aggregated signals indicate illicit activity is more likely than not, rather than waiting for airtight proof. In practice, that standard favors systems that can aggregate patterns, spot anomalies, and connect disparate signals faster than any manual investigation — exactly the kind of connective tissue that AI-driven systems can surface better than human investigators.

Terrorist finance: tip of the spear

Terrorist finance is likely to feel the impact of this enforcement shift before other domains. The reason is twofold. First, national security considerations ensure that counter-terrorism receives the highest priority for resources, political will, and legal innovation. Governments simply cannot afford to take a passive stance when designated groups are exploiting new technologies.

Second, counter-terrorism already has over two-decades-worth of laws, regulations, and intelligence cooperation to build upon — from FATF standards and UN Security Council resolutions to domestic tools like the U.S. Patriot Act and equivalent frameworks worldwide. That foundation makes terrorist finance the most immediate proving ground for applying aggregated data and AI-enabled detection under a preponderance standard, before those same methods expand into broader areas of illicit finance.

Conclusion

Progress in this field isn’t measured by perfection — it’s measured by whether the system gets harder for bad actors to exploit. And by every sign — data aggregation, AI adoption, and regulatory convergence — we’re moving in that direction.

Caution is still warranted; threat finance adapts. But the fatalism that says this space is uncontrollable no longer matches reality. On this issue at least, I am cautiously optimistic.

Note: This article is in part a preview of a bigger piece I’m currently working out. Also, stay tuned for the long-awaited Part II of my China military series.