Cobo outlines how AI agents could reshape stablecoin payments and wallet security

Cobo outlines how AI agents could reshape stablecoin payments and wallet security

N
News Editor
2026-10-21 10:35:54
Cobo says AI agents are moving beyond information processing and into autonomous execution, with payments becoming part of that shift. Speaking at the fourth Cross-border Digital Finance Compliance Conference hosted by Mobile Payment Network on Oct. 15, Cobo head of sales and solutions Lucas discussed how AI Agent use cases are changing payment patterns, why stablecoins and low-cost blockchains are being tested for high-frequency micropayments, and what security controls are needed when software starts initiating transactions on its own. The presentation covered several layers. On the demand side, Lucas pointed to AI-driven API calls, data purchases and compute spending, where transaction frequency may rise while ticket sizes shrink to just a few cents or less than $0.01. On the security side, he said AI wallets need more than private key protection, highlighting MPC-based key management, policy constraints and pre-signature verification as ways to separate autonomous execution from control over funds. He also described the infrastructure needed to bring stablecoin payments into enterprise workflows, including collection, KYT-based screening, bookkeeping, payouts and settlement. For traditional payment firms, Cobo presented a modular middle path between relying fully on third-party providers and building a full on-chain stack from scratch.

AI agents are moving from information processing to autonomous execution, and payments are starting to become part of that shift.

Cobo outlines how AI agents could reshape stablecoin payments and wallet security 2

On Oct. 15, at the fourth Cross-border Digital Finance Compliance Conference hosted by Mobile Payment Network, Lucas, head of sales and solutions at Cobo, gave a presentation on the integration of AI agents and Web3 stablecoin payments, with a focus on security and risk controls. Drawing on industry discussions held earlier during TOKEN2049 in Singapore, Lucas laid out Cobo’s view on the rise of AI agent payments, the security boundaries of agentic wallets, the core pieces of stablecoin payment infrastructure, and how traditional payment institutions can enter the Web3 payments market through modular infrastructure.

On-chain AI payments are emerging as billing shifts from subscriptions to per-use charges

As AI agents begin calling APIs on their own and purchasing data or computing resources, payment demand is changing with them. Transactions may become more frequent, while the amount per transaction gets smaller.

According to the article, traditional payment rails struggle to handle this type of activity economically. Credit cards and similar channels usually carry fixed fees. For an API call worth only a few cents, or even less than $0.01, the payment cost can exceed the price of the service itself. Low-cost blockchain networks and stablecoins offer another settlement option for these high-frequency, low-value transactions.

That pattern is already showing up on-chain. Citing a sample study covering more than 6 million transactions over 35 days on Base and Solana, the article says more than 90% of AI-related payment transactions in the sample were only a few cents or less than $0.01. In Cobo’s telling, machine payments are starting to show a value distribution that looks very different from traditional consumer payments.

The clearest early use cases are in APIs and online services. On Solana, the article points to a weather query API priced at less than $0.01 per call and a token security screening service that charges a few cents each time. Services that once relied on subscriptions or prepaid balances are beginning to test per-call billing, allowing AI agents to buy resources only when needed.

Similar experiments are extending to internet content access. The article notes that companies including Cloudflare are pushing pay-per-use mechanisms based on the x402 protocol for scenarios such as AI crawlers accessing website content and automated service calls.

For enterprises, the potential use goes beyond API micropayments. Cobo said AI agents could initiate batch payments within preset budgets and permission ranges in areas such as overseas supplier settlement, advertising delivery and channel commission payouts, while pairing those flows with automated reconciliation, abnormal transaction detection and on-chain AML screening to reduce manual work.

These early cases suggest that payments are expanding from human-initiated actions to transactions launched by software in response to business needs. As frequency rises and payment sizes keep fragmenting, infrastructure needs to support not only low-cost settlement but also automated execution and the controls that go with it.

Agentic wallet security is not only about private keys but also about transaction boundaries

Once AI agents begin operating wallets on their own, the security problem becomes more complicated.

The article says that, unlike a traditional wallet, an AI agent’s actions may be affected by model hallucinations or malicious prompt injection. Even if the private key itself is not exposed, the agent could still initiate transactions that go beyond the user’s actual intent because of faulty instructions or outside manipulation.

Cobo outlines how AI agents could reshape stablecoin payments and wallet security 3

That changes the security target for AI wallets. Protecting the private key is no longer enough on its own. The wallet also needs clear permission boundaries for autonomous behavior.

To address that, Cobo said its Agentic Wallet uses MPC, or multi-party computation, for key management, Pact policy constraints and pre-signature verification to control agent behavior at different stages.

  • MPC key management: This is meant to keep AI from controlling funds on its own. With MPC, transaction signing is completed jointly by multiple independent parties, so the AI agent does not hold the full private key. Even if the agent’s runtime environment is compromised, an attacker would have difficulty completing a signature using only the key material available in that environment.
  • Pact policy constraints: These rules define the agent’s operating range in advance. Enterprises or wallet owners can authorize payment rules ahead of time, including approved counterparties, per-transaction and cumulative limits, validity periods and trigger conditions. The agent can act autonomously within those rules, but transactions outside them cannot directly obtain signing approval.
  • Pre-signature verification: This acts as the final checkpoint before funds leave the wallet. The system can set tiered approvals based on transaction amount, recipient address and business rules. Small transactions that fit established policies can be executed automatically, while transactions above a set threshold or those that trigger risk conditions require human confirmation. By independently checking permissions at the signing stage, the system reduces the risk created by tampered front-end instructions or abnormal agent behavior.

The core idea is to separate an AI agent’s ability to execute from actual control over funds, allowing the agent to complete authorized payment tasks while enforcing restrictions on critical transactions in the signing flow.

Stablecoin payments need more than a wallet to fit into enterprise operations

An agentic wallet answers one question: within what permission range can AI use funds? But for stablecoin payments to move into enterprise workflows, the article argues that companies also need infrastructure that can handle collection, fund screening, bookkeeping, payouts and settlement.

Lucas grouped the main challenges for traditional payment institutions into three areas: how to connect stablecoins to existing technology and licensing structures, how to identify on-chain fund risks, and how to make sure each transaction is executed as intended.

On system architecture, Lucas described a 2×2 matrix for stablecoin payment integration. At the asset management layer, a company can choose third-party custody or self-custody. At the system integration layer, it can choose a standardized payment page or embed the function into an existing payment system through APIs or SDKs.

Those choices can be combined based on business needs, technical capacity and compliance requirements. For institutions that want to keep their current payment gateway, a modular integration model means they do not need to build a full on-chain stack from scratch.

Compliance screening is another key piece. Unlike card payments, an on-chain stablecoin transfer usually cannot be reversed through a chargeback mechanism once it is confirmed. That means companies need to identify high-risk addresses, unusual fund flows and potential AML risks during both incoming and outgoing payments.

Cobo said its approach combines data from KYT, or know-your-transaction, partners with AI-assisted analysis to provide compliance teams with transaction risk alerts and fund flow information, supporting later review and handling decisions.

The article also highlights transaction execution security as an area that is often underestimated. Based on digital asset security incidents in recent years, attackers do not always need to steal a private key directly. If a front-end page, business system or third-party service is compromised, transaction instructions can still be altered, causing a legitimate signing process to send the wrong payment.

To address that risk, Cobo said it uses a layered defense model across three stages: controlling access and call permissions at the service request layer, constraining transactions through order creation, review and multi-level approvals at the business process layer, and independently verifying transaction intent and authorization rules at the final signing layer.

Cobo outlines how AI agents could reshape stablecoin payments and wallet security 4

In the article’s framing, that design matters because even if an upstream business system is attacked, abnormal transactions may still be identified and blocked before the final signature, rather than relying entirely on the security of the front end.

A modular middle path for traditional payment firms entering Web3

For traditional payment institutions looking to expand into stablecoin business, entering Web3 is first a trade-off between technical investment and control over operations.

At one end, a firm can rely on a third-party provider for full payment and asset management capabilities. That is easier to integrate, but control over the underlying system and fund flows may be limited. At the other end, a firm can build its own infrastructure for on-chain acquiring, fund collection, address management and gas fee management. That offers more autonomy, but it also brings ongoing development and maintenance costs.

Based on its work with a licensed payment institution in Hong Kong, Cobo said it has been testing a third route between those two options: modular payment middleware that lets enterprises integrate existing on-chain payment capabilities directly while choosing asset management and system integration methods according to their own business and compliance needs.

Under that model, payment institutions do not need to repeatedly build common blockchain functions. They can put more resources into customer service, business expansion and compliance operations, while retaining control over assets and key business processes through the right architectural setup.

For traditional payment firms, the article says modular infrastructure offers a more flexible way into Web3. It allows them to combine underlying technical capabilities with their own business systems instead of choosing only between full dependence on outside providers and a fully self-built stack.

Cobo’s current direction

The article closes by saying AI agents are bringing new use cases to on-chain payments, though current applications are still concentrated in small-value, per-use and rules-based automated transactions.

From agentic wallets to stablecoin payment infrastructure, one direction is becoming clearer in Cobo’s view: as payment execution becomes more automated, enterprises need to place fund control, transaction authorization and risk management on mechanisms that are more independent and more verifiable.

Cobo said this will remain part of its ongoing work, using modular asset security and payment infrastructure to provide composable building blocks for AI agents, traditional payment institutions and Web3 developers.

Disclaimer: Markets involve risk, and investment requires caution. This article does not constitute investment advice. Users should consider whether any opinion, view or conclusion in the article fits their own circumstances. Any investment decision made on that basis is at the user’s own risk.

This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan.
8900

Disclaimer:

The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.

Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.

Cobo outlines how AI agents could reshape stablecoin payments and wallet security | Bit.Fan