Money20/20 Vegas 2026: Sumsub's Vyacheslav Zholudev on the rise of AI agents

Quick Summary

As the industry enters a new technological revolution, many companies remain unprepared for the risks associated with autonomous AI. Vyacheslav Zholudev of Sumsub explains at Money20/20 Vegas 2026 how the concept of Know Your Agent (KYA) is becoming essential to prevent hallucinations, trickery, and irreversible harmful actions in the agentic era.

What are the primary risks of autonomous AI agents?

The industry is currently not ready for the shift toward agentic AI, and there is a high probability that many firms will face significant challenges before adapting. The most obvious risks involve agents performing unintended actions, such as hallucinating or being manipulated by users into performing harmful, irreversible tasks. Without proper controls, these automated systems can quickly move in a direction that causes damage to the business or its customers.

How does Sumsub define Know Your Agent (KYA)?

KYA is a broad term that covers several distinct challenges in the current landscape:

• Detection: Identifying whether an API call or a website visitor is a human or an automated agent.
• Identity Attachment: Determining how to link a specific identity to an agent, particularly in scenarios like anonymous agents attempting to purchase cryptocurrency.
• Control: Monitoring and managing what an agent is permitted to do, especially when accessing sensitive internal documents or performing financial tasks.

How can companies control agent executions?

Sumsub has recently utilised its transaction monitoring solution to oversee MCP tool executions. By treating an agent's action as a transaction, companies can apply specific rules to block certain executions or trigger additional security measures, such as a liveness test, when a high-risk action is detected. This allows for a balance between automation and safety, as requiring human intervention for every single action would defeat the purpose of using AI agents in the first place.

What is the current state of regulation for AI agents?

The current environment is described as a bit of a Wild West. Significant problems arise around legitimacy and accountability, such as when a user claims a transaction made by their agent was unauthorised to trigger a chargeback. Regulators must focus on three core areas: protecting user data, controlling financial operations, and distinguishing legitimate automation from illegal activities. As agents become more autonomous and perform increasingly complex tasks, verifying their actions will likely mirror modern transaction monitoring.

FF NEWS TAKE:

The transition from KYC to KYA marks a pivotal moment in the evolution of digital compliance. By repurposing transaction monitoring tools to oversee AI agent executions, Sumsub is providing a pragmatic path forward for firms that want to embrace automation without sacrificing security. As these agents move from simple tasks to writing code and managing finances, the ability to treat every automated action as a verifiable transaction will be the only way to maintain trust in the financial ecosystem.