• Madhu Nadig, Co-founder & CTO at Flagright


• 27.04.2026 03:45 pm
#Compliance #FinancialCrime

The latest INTERPOL threat assessment makes the uncomfortable point that modern financial fraud no longer looks like a collection of isolated scams – it looks like an industry. INTERPOL ranks financial fraud among the top global crime threats, estimates 2025 global losses at roughly $442 billion and describes fraud as sitting at the centre of “polycriminality,” intersecting with organised crime, human trafficking and cybercrime. 

Scam centres are no longer a regional anomaly either. INTERPOL says they have spread globally, involve hundreds of thousands of people and have trafficked victims from nearly 80 countries into forced online fraud.

That matters because organised fraud networks increasingly behave like full-service businesses. INTERPOL’s 2026 assessment says criminal peer-to-peer marketplaces now act as full-service hubs for fraud syndicates, offering phishing tools, fake trading platforms, AI-powered chatbots for victim grooming, encrypted communications and integrated money laundering support. 

The same report says offenders are adapting their business models for efficiency by collaborating with specialised money laundering networks. In other words, fraud groups are no longer improvising but are assembling modular supply chains where one team handles outreach, another persuasion, another credential theft, another mule recruitment and another laundering and cash-out. Compliance teams, by contrast, still often investigate one alert, one account, one queue at a time.

Artificial intelligence has made that industrialisation more powerful, not because it invented fraud, but because it has compressed the cost and time of persuasion. INTERPOL says AI-enhanced fraud is 4.5 times more profitable than non-enhanced fraud and warns that agentic AI can autonomously support large parts of a fraud campaign. It also highlights how voice and face cloning tools can be built from seconds of real audio or video. The FBI’s 2025 Internet Crime Report shows how this is already showing up in the real world. The Internet Crime Complaint Center logged more than 22,000 complaints referencing AI, with reported losses approaching $893 million, and the FBI specifically noted scammers’ use of fake social profiles, voice clones, fabricated IDs and believable videos.

You can see the effect most clearly in so-called “pig butchering” scams, which regulators increasingly describe more formally as relationship investment scams. The Commodity Futures Trading Commission says these schemes use dating apps, social media, messaging apps and even ‘wrong number’ texts to initiate contact, then rely on fake profiles, images, videos and voices to build trust before moving victims onto fraudulent trading platforms. FinCEN has similarly warned that these scams are often run by overseas criminal enterprises that use victims of labor trafficking to conduct mass outreach. And the FBI’s 2025 report shows how profitable this model has become as cryptocurrency investment fraud generated $7.2 billion in reported losses in 2025, while account takeover complaints reached roughly 4,700 with about $359.7 million in losses. The important shift is that AI helps criminals personalise scams at scale while keeping the unit economics attractive.

The laundering side is just as networked. The joint FATF-INTERPOL-Egmont report on illicit financial flows from cyber-enabled fraud describes syndicates structured into specialist sub-groups, including money laundering teams, and notes that laundering networks can involve money mules, shell companies, legitimate businesses, banks, payment and remittance providers and virtual asset service providers. The proceeds are often moved rapidly through networks of accounts spanning multiple institutions and jurisdictions. 

The same report says criminals use social media and messaging platforms to recruit money mules across borders at scale. FinCEN has separately warned that scammers increasingly direct victims to convertible virtual currency kiosks, where cash can be converted into crypto quickly; in 2024 alone, the FBI received nearly 11,000 complaints involving these kiosks, with reported losses of about $246.7 million. FATF’s latest work on stablecoins and unhosted wallets adds another layer, noting that threat actors use stablecoins, mixers, cross-chain bridges, peer-to-peer channels and OTC brokers to complicate tracing and exit into fiat.

This is why so many compliance teams are struggling. The problem is not simply that criminals are moving faster but that many organisations still respond through fragmented structures built for an earlier era, covering fraud in one team, AML in another, sanctions somewhere else, cyber in another function, and investigations spread across separate tools and data stores. 

The FATF-INTERPOL-Egmont report says jurisdictions need to analyse ‘voluminous information inflows’ and build strong cross-cutting coordination mechanisms because cyber-enabled fraud is not a single-category problem. INTERPOL makes the same point bluntly in its recommendations. Authorities and institutions need to treat fraud as a networked, transnational threat rather than a series of isolated incidents. From what we see at Flagright, the operational bottleneck is rarely the first alert. It is the inability to connect identities, accounts, devices, wallets, counterparties and payment behavior quickly enough to understand the network behind the alert.

So what should organisations do differently? First, they need to stop treating fraud controls and AML controls as separate operating systems. The FATF-INTERPOL-Egmont report explicitly says anti-fraud and AML processes are complementary and points to real-time transaction monitoring as a useful control for identifying and preventing fraudulent or illicit activity. 

Second, they need to move from transaction-centric detection to entity- and network-centric detection. The question is not only whether a single payment looks suspicious. It is whether the broader pattern across customers, accounts, wallets, devices, merchants and destinations looks coordinated. 

Third, they need to use AI in the right place. The biggest opportunity is not just generating more alerts. It is compressing investigation time, including gathering context, checking prior internal history, identifying linked entities and standardising evidence collection so humans can focus on ambiguous and higher-risk cases. But that only works if the AI is bounded, auditable and easy to challenge.

Finally, firms need stronger intervention muscle, not just better dashboards. Criminal networks exploit the fact that funds can move across institutions and borders almost instantly. The joint FATF-INTERPOL-Egmont report stresses that proceeds are often laundered across multiple jurisdictions and institutions and that rapid multilateral cooperation is essential to intercept them. 

INTERPOL’s own recommendations call for stronger intelligence sharing, better suspicious transaction information exchange, improved oversight of virtual asset service providers and more coordinated mapping of the criminal networks behind fraud. For firms, that means having playbooks for mule-account disruption, cross-channel case management, urgent payment holds, law-enforcement escalation and rapid sharing of suspicious platform and wallet intelligence.

The core mistake is to think this is merely a volume problem. It is an architecture problem. Criminals have already reorganised around networks, specialisation, speed and channel-hopping. Many compliance teams are still organised around alerts, silos and delayed handoffs. That mismatch is now expensive. The institutions that will close it are the ones that unify fraud and AML thinking, investigate behavior as networks rather than isolated events and use AI not as a branding layer, but as an operational tool to make detection, investigation and intervention faster and more defensible.