By Pierre Gérard, CEO and co-founder, Scorechain

When Scorechain was founded in Luxembourg in 2015, "blockchain analytics" had not yet emerged as a recognized category. The early years were spent explaining to banks and regulators why the transparency of a public ledger represented an opportunity rather than a threat. A decade later, the same misunderstanding is now attaching itself to artificial intelligence (AI), and it is costing the industry valuable time.

Two narratives dominate the current conversation. The first claims that AI will soon replace compliance teams entirely. The second argues that AI is too unpredictable to be trusted near regulated financial activity. Neither holds up to scrutiny. Since 2015, Scorechain has risk-assessed more than 2,800 virtual asset service providers (VASPs), and over the past two years the company has integrated AI where it delivers genuine value — as a separate layer rather than something embedded directly into the compliance tools clients rely on.

The Noise Problem in Compliance

Every compliance officer raises the same issue within the first five minutes of conversation: noise. A sanctions screening system tuned to the level a cautious bank prefers can generate false positives at rates around 95%. Transaction monitoring produces similarly high rates. The consequence is that trained analysts — professionals who understand typologies and can interpret fund flows — spend most of their workday clearing alerts that never represented real risk: dismissing name matches on common surnames, or reading through multiple adverse media hits that turn out to describe entirely different individuals. Each adverse media check typically consumes 10 to 20 minutes of an analyst's time. This is the daily reality of compliance work, and it is a primary reason skilled professionals leave the field. Industry associations such as ACAMS have flagged rising burnout and attrition in anti-money-laundering roles as a structural problem, not a cyclical one.

This is where automation proves its worth, though in a narrower capacity than the hype suggests. Scorechain AI is designed to deliver a consolidated report to the analyst: a wallet's risk score, the entity types it has interacted with, and the named services and counterparties it has been exposed to. This represents work that previously required hours of manual tracing across a ledger. Critically, the report does not make decisions. It compresses the evidence so that the compliance officer — the person responsible for signing off on the decision and defending it to a regulator — can review it in minutes and make the final call. Effective automation does not shrink the compliance function; it shifts the focus from repetitive triage back to professional judgment.

Why Human Accountability Is Architectural, Not Temporary

That distinction matters because the alternative carries real danger. In a regulated environment, a model cannot answer to a supervisor. Both the Sixth Anti-Money Laundering Directive (AMLD6) and the Markets in Crypto-Assets Regulation (MiCA) require institutions to explain and stand behind their decisions. MiCA, which began taking effect across the European Union in mid-2024 with provisions phasing in through 2025, imposes some of the most detailed operational obligations yet applied to crypto-asset service providers, including requirements around complaint handling, prudential safeguards, and market abuse monitoring. "The algorithm flagged it" does not constitute a valid defense during an inspection, and "the algorithm cleared it" is even worse. The only responsible design approach is AI functioning as a support layer built on top of trustworthy data, with a named compliance officer retaining both the decision-making authority and the accountability that accompanies it. Human oversight is not a temporary measure to be phased out as models mature — it is the architecture itself.

A model's quality is ultimately determined by what sits beneath it. This is what outsiders tend to miss. On its own, an AI system reading a blockchain perceives only anonymous strings of characters transferring value to other anonymous strings. It cannot identify that a wallet three hops upstream belongs to a sanctioned exchange, or that the counterparty receiving funds is a mixer rather than a payroll provider. Supplying that missing context is the core function of blockchain analytics: attaching identity and risk assessments to raw on-chain activity, tracing indirect exposure across multiple hops rather than merely checking the address in front of you, and scoring it against more than a billion data points and over a million crypto entities that Scorechain has labeled since 2015.

Consider a concrete example: a wallet appears clean on initial inspection, but tracing its flows reveals that most of its balance originated two hops back from an address tied to a ransomware operator. That is a finding no model could reach on raw chain data alone, yet it is exactly the type of finding a compliance officer must act upon. Providing a model with that context allows it to reason on solid ground. Feeding it thin data produces confident nonsense — which in compliance is more dangerous than an honest gap, because it results in clearing transactions that should be flagged.

AI as Market Participant, Not Just Compliance Tool

A genuinely new development is that AI is no longer solely a tool used by compliance teams; it is becoming a participant in the market they monitor. Autonomous agents that initiate payments under preset limits have moved from demonstration to deployment, driven by real infrastructure: Coinbase's x402 standard for machine-to-machine payments, Visa's Trusted Agent Protocol, and the PayPal and OpenAI checkout integration. Software is beginning to transact with other software, settling in crypto assets at volumes no treasury team could process manually. This is arriving at a time when more than 50 jurisdictions have already adopted or proposed FATF Travel Rule requirements for VASPs, meaning the compliance perimeter for identifying counterparties is already expanding beyond what manual processes can sustain.

This raises a question the industry has not yet answered clearly: how do you apply Know Your Transaction principles to a counterparty that is a piece of software? The direction, at least, is apparent. When agents transact independently, controls cannot reside solely at the onboarding stage. They must move to the transaction layer itself — real-time monitoring, velocity limits, provenance tracking, and the ability to intervene while funds are still in motion. The transparency that was defended to skeptics for years turns out to be the one characteristic that makes autonomous on-chain activity auditable at all.

Building for Autonomous Compliance

Scorechain recently launched Scorechain MCP, which exposes its risk scoring and entity intelligence through the Model Context Protocol — an emerging standard that enables AI agents to call external tools directly. The intelligence resides within the Scorechain platform, while the AI remains outside it, calling in for answers rather than being embedded in the compliance tool itself. The premise is straightforward: an agent should never transact without information. Before moving funds or approving a counterparty, an agent can query Scorechain to determine whether an address belongs to a sanctioned entity, a mixer, a known scam, or a clean private wallet, and receive a risk score in return.

This is not hypothetical capacity. Scorechain currently processes more than 1.5 million AML checks per day, with a screening call returning in approximately 235 milliseconds — fast enough to operate inside a live transaction without causing delay. The service is available where agents and workflows actually operate: as an app inside ChatGPT and Claude, and as an integration on automation platforms such as n8n and Zapier. A compliance check embedded in the flow at the moment of decision carries far more value than one attached after funds have already moved.

The most fundamental check of all — whether an address appears on a sanctions list — should not be gated behind a paywall. For that reason, Scorechain offers a free sanctions screening API that any developer, agent, or workflow can call. Screening for sanctions exposure is not where a compliance provider should extract value; it represents the baseline the entire market should stand on.

Asset-Level Risk Intelligence

A second-order shift is emerging that token issuers and asset managers are only beginning to confront. When value moves into stablecoins and tokenised assets at machine speed, the risk that matters extends beyond the individual transaction to the asset itself: who holds it, how concentrated that ownership is, and how much of the supply resides with sanctioned or high-risk entities. This is a fundamentally different question from transaction monitoring, and it is the one Scorechain's Digital Asset Intelligence is built to address — providing issuers and asset managers with an asset-level view of holders and exposure before they mint, list, or allocate. The urgency is growing as institutional engagement with tokenised assets accelerates, with firms such as BlackRock launching tokenised treasury funds on public blockchains and major banks piloting settlement infrastructure using stablecoins.

Europe is better prepared for this transition than it is generally given credit for. MiCA and AMLD6 already assume continuous monitoring and clear accountability rather than a one-time check at onboarding. A regulatory regime that requires activity to be explainable is precisely what is needed when software begins moving money autonomously.

AI is indeed transforming crypto compliance. It will eliminate a significant amount of tedious work, which is a welcome development. What it will not eliminate is the need for judgment, accountability, and verifiable data. If anything, it raises the standard on all three. The teams that approach AI as a faster analyst — one grounded in reliable on-chain intelligence and kept firmly under human control — are the ones positioned to remain effective when machines begin transacting on their own. That future is closer than most people think.