Banks and fintechs are adding customers at record speed. That’s good for growth—and hard on compliance. Alert volumes spike, onboarding queues swell, and the old mix of batch screening and manual casework starts to creak. If you’ve felt this inside your own program, you’re not alone. Across the industry, institutions are discovering that scaling AML without sacrificing accuracy now depends on a new operating model: AI-assisted controls, continuous monitoring, and a unified view of customer and on-chain risk.
The shift isn’t optional. Global bodies are raising the bar, and new supervisors are turning guidance into day-to-day oversight. FATF’s June 2025 targeted update again pressed countries to close gaps in their virtual asset and VASP regimes, noting uneven implementation and calling for stronger supervision and enforcement. In the EU, the new Anti-Money Laundering Authority (AMLA) in Frankfurt has begun operations and will ramp up staffing and IT through 2026–2027 before assuming direct supervision in 2028—clear evidence that European AML expectations are consolidating and intensifying. At the same time, MiCA’s staged application is pushing crypto-asset service providers to professionalize risk management and licensing, with stablecoin rules in force since June 2024 and broader CASP obligations phasing in from late 2024 into 2025.
Against that backdrop, one truth explains much of the urgency: traditional AML setups generate far too many false alarms. Peer-reviewed research and industry analyses still cite 90–95% false-positive rates for legacy transaction monitoring—a drain on analyst time and a real reason genuine risk can be missed. Add in deepfake-driven identity fraud—rising through 2024–2025—and the pressure on onboarding and KYC is obvious. So what does “good” look like now? And how can AI and RegTech help you scale without letting quality slip?
From point tools to a single risk spine
Most compliance pain comes from fragmentation: one vendor for KYC, another for sanctions, a third for adverse media, a fourth for on-chain analytics, a home-grown rules engine, and a case tool that only sees half the data. Analysts spend their day stitching together a story rather than making a decision.
The fix is architectural as much as analytical. Move toward a single risk spine that every control reads from and writes to:
AI belongs at each layer—but only if the data is clean, the outcomes are explainable, and humans can override when needed.
Where KYC-Chain fits: many institutions use KYC-Chain to anchor identity and screening—document authentication, biometric liveness, sanctions/PEP/adverse-media checks, and configurable decisions—while feeding outcomes to monitoring and case systems. That way, you don’t trade speed for auditability; you keep both.
Onboarding that blocks synthetic IDs without punishing good users
The volume of synthetic identities and deepfake attempts is rising, fueled by cheap generative tools. Several industry sources report double-digit growth in deepfake attacks into 2025, targeting both onboarding and step-up authentication.
Practical moves that work in the wild:
KYC-Chain’s document + liveness stack is designed to live inside that flow and keep the decision trail intact—what was checked, when, and why approval (or a referral) was made. If you’re facing board questions about deepfake resilience, that evidence matters as much as the model.
Sanctions, PEP, and adverse media as a living control
Screening isn’t a one-time lookup. Lists change; names have aliases and transliterations; PEP risk varies by role and region. Supervisors expect continuous screening with verifiable list versions, and they ask to see how analysts dispositioned edge cases.
Good teams treat screening as a living radar: feed updates daily (or faster), enrich with secondary identifiers (DOB, nationality) to reduce false hits, and route meaningful matches to enhanced due diligence rather than defaulting to blanket denials. This pays off twice: fewer false positives up front, and stronger narratives when you do file.
Transaction monitoring that finds signal—and proves it
Rules versus AI isn’t a choice; it’s a blend. Rules cover crisp policy triggers—sanctions hits, unusual first withdrawals, velocity spikes post-deposit. AI/ML picks up the long tail—behavior that deviates from a customer’s baseline, clusters of addresses moving in lockstep, patterns that stretch across assets and chains.
To keep the balance right:
External data underscores the need to keep tuning. Research states that $40.9B in 2024 value flowed to addresses it currently classifies as illicit, and notes continued professionalization and diversification of criminal methods—signals that static rulebooks age quickly.
On-chain analytics: beyond simple “taint”
If your on-chain policy stops at “one hop from a bad cluster,” you’ll over-block low-risk flows and still miss crafty ones. Entity attribution and path context are where detection quality lives.
A deposit address for a major exchange is not the same as a darknet market hub. A first-hop exposure to a sanctioned mixer is different from a far, time-diluted exposure with high volumes of clean flow in between. DeFi adds its own wrinkles—fresh proxy contracts, exploit-tagged addresses, and bridges used to layer funds quickly.
Modern policy reflects this with stepped friction: add cool-off windows and step-up checks for first-time withdrawals, new devices, or brand-new withdrawal addresses; scale limits up with clean, consistent behavior; scale down when counterparties raise risk. EU authorities building AMLA, and FATF’s June 2025 update calling for stronger VASP oversight, both point to a world where this kind of nuance is expected—not “nice to have.”
Perpetual KYC (pKYC): the end of “set and forget”
Risk changes when life changes. Customers move countries, change jobs, switch devices, or add new payment methods. pKYC replaces the “every X years” refresh with event-driven reviews triggered by behavior and profile changes: a new device on an old account, a sudden jump in stablecoin volume, a large cross-border withdrawal, a new legal representative on a business account.
AI helps decide when a nudge becomes a re-verification and what evidence is proportionate. The result is less customer friction at low risk and faster escalations when risk actually rises.
Governance for AI-assisted AML
AI is not a free pass; it adds responsibilities. Three disciplines differentiate high-maturity programs:
Industry analysts—including Gartner’s reviews of AML tools—consistently point to case-investigation AI, automated triage, and straight-through processing (STP) for low-risk alerts as levers that both cut cost and raise quality when wrapped in strong governance.
Regulatory momentum: what to brief your board on this quarter
If you operate globally, remember that local enforcement continues to cite foundational lapses: backlogs in onboarding and alert queues, sanctions misses, and under-resourced programs. The headlines may differ, but the fixes are the same.
What changes on Monday morning
Start where the pain is most visible. If analysts are overwhelmed, measure your false-positive rate and time-to-disposition, then fix the causes: deduplicate identical alerts, merge related ones into a single case, and re-weight rules that fire constantly without producing SARs. If onboarding is slow and leaky, add biometric liveness and device checks and tie them to step-up flows rather than front-loading friction for everyone.
Wherever you begin, connect outcomes back to a single customer risk score that drives limits, step-ups, and case priority. That’s the spine your controls need.
KYC-Chain’s role in this picture is straightforward: give your first line a fast, reliable way to establish identity and risk—documents, liveness, sanctions/PEP/adverse media, and clear decisions with a full audit trail—so monitoring and investigators start from ground truth, not guesswork. As you add AI and on-chain analytics around that core, you scale without losing clarity.
Closing thought
AML didn’t suddenly become harder; financial crime evolved and the rulebook caught up. AI and RegTech aren’t shortcuts—they’re how modern teams keep pace with growth, give customers a smooth experience, and still deliver the evidence regulators expect. With a joined-up stack and a few disciplined habits, you can have both speed and certainty.
Want to see how real-time AML monitoring, sanctions screening, and AI-powered KYC work together in one platform? Book a demo with KYC-Chain and experience the future of compliance at scale.
Identity Verification in Daily Life: How You Prove Who You Are Online
The Importance of Audit Trails in Enterprise Compliance
AI Compliance Agents for KYC/AML in 2026: Hype vs. Reality