AI Potential for Banks and Payment Systems Compliance Process

Undergoing compliance check in payment system is one of the most painful procedures for almost any business entering new market or seeking additional providers. Endless document packages, clarification emails, weeks of waiting, and in the end — either a rejection without explanation or a request for yet another set of papers. Sound familiar?

Like many other members of fintech industry, we’ve learned to navigate this: preparing documents in advance, finding the “right” people inside, and anticipating requests based on past experience. Many businesses, to be honest, rely on sole hope that compliance officer will eventually get tired and overlook minor inconsistencies just to close the case faster.

This strategy worked as long as humans were on the other side. But since late last year and early this year, the approach to compliance and KYB (Know Your Business) procedures has changed dramatically. Artificial intelligence has not bypassed this one, and many financial companies have already integrated it into their systems. And we’re not just talking about assisting professional — this is a fundamentally new approach to verification, where AI sees right through everything.

Goldman Sachs and the New KYB Standard

In February 2026, Goldman Sachs announced that over the past six months, in collaboration with Anthropic, it had been developing AI agents for client onboarding and KYC (Know Your Customer) processes. These aren’t just chatbots for discussion — they’re autonomous agents capable of verifying documents, extracting data, and identifying inconsistencies. Based on this data, Claude (Anthropic AI) will analyze company ownership structures and flag borderline cases for final human review. Testing has shown a 30% increase in onboarding speed. Marco Argenti, the bank’s CIO, describes Claude as a “digital colleague for professions where work is large-scale, complex, and highly process-dependent.” Ultimately, humans won’t have to spend hours cross-checking data — they’ll receive a concise summary to make an accurate decision. This process is the most labor-intensive and error-concentrated, as the human brain can’t process such vast amounts of information at once the way AI can.

Another major bank, JPMorgan, predicts that AI will make its KYC processes 90% more efficient, allowing it to process hundreds of thousands of documents with significantly fewer staff than it currently employs. HSBC has implemented systems that detect 2-4 times more suspicious transactions while reducing false positives by 60%. Payment systems aren’t lagging behind either. Stripe is already tightening KYC and AML requirements to maintain active accounts. Marketplaces are rejecting sellers who can’t pass basic compliance checks. Pressure is coming from all sides of the ecosystem, and tech providers are becoming part of this wave.

These examples prove that such implementations will continue to evolve. The main advantage of AI is its meticulousness and absence of human bias. For AI, everyone is equal, and data is compared in real time so quickly and accurately that errors are nearly impossible.

What Exactly Does AI Do in Compliance?

It’s important to understand that we’re not talking about a scanner that knows everything about you. AI agents operate within specific layers: they read and interpret documents, cross-check data with each other and external registries, apply rules, and — crucially — handle cases where there’s no clear rule but logical judgment is required. This is where traditional automation always failed: if a passport was scanned at a wrong angle or an address was written in a non-standard format, the system would wait for human intervention. An AI agent, in most cases, will figure it out on its own and interpret the data correctly.

At the same time, there’s a shift toward continuous KYC rather than one-time checks during onboarding. A client’s profile can now be updated automatically based on changes in registries, news publications, or transactional behavior. If a company’s ownership structure changes or a name appears on a sanctions list, the system will react immediately — no need to wait for the next scheduled review in six months. While regulators have long expected companies to continuously monitor their clients against sanctions lists, doing this manually is impractical for most businesses. Processing a constant stream of information for all clients daily can only be done through automation, and AI handles this better than anything else.

Will it Become Simple or Complex? — Depends on Your Business

For businesses with transparent structures, clear business models, and clean documentation, automation is actually the good news. The process will be faster, and the outcome more predictable. The human factor will be less involved in procedures, and an employee’s mood on a given morning won’t affect the decision on a merchant.

For businesses with non-standard structures, holdings across multiple jurisdictions, or unconventional business models, things may become more complicated. When faced with uncertainty, an AI agent doesn’t turn a blind eye or “cut slack” — it escalates, requests additional documents, and demands explanations. Complex cases aren’t expedited; they become more demanding in terms of preparation quality.

In response to these changes, regulators require that every automated decision be explainable. This means banks are building systems where every verification step is documented. On one hand, this protects the client: if rejected, there must be a precise reason (though it won’t always be disclosed). On the other hand, the bar for document quality and business model logic is rising. Simply saying “we sell digital goods” is no longer enough — you’ll need to describe your activities in detail, understanding that they’ll be run through a logic system that might not know everything.

What Does This Mean in Practice?

Ultimately, the days when you could rely on a compliance officer’s fatigue are coming to an end. AI doesn’t get tired, doesn’t rush, and doesn’t overlook a field in an Excel table. This doesn’t mean legitimate businesses won’t pass onboarding, but preparing for it is now a separate task that can’t be left to chance.

For businesses with simple models, AI-driven compliance is more of a simplification, minimizing interaction with compliance officers. However, compliance-intensive businesses (including iGaming, dating, and subscription services) will face even more challenges, where every mistake could cost them their merchant account.