When AI agents begin making autonomous payments: Cobo Lucas discusses stablecoin payments and security infrastructure.

As AI agents accelerate their move towards autonomous execution, how can payment infrastructure adapt to new application demands?

AI agents are moving from information processing to autonomous execution, and payment is beginning to become a part of this process.

When AI can autonomously call APIs, purchase computing resources, or complete on-chain transactions, the initiator of payments, transaction frequency, and amount structure may all change. In particular, high-frequency, small-amount machine-to-machine payments are posing new requirements on the cost structure of traditional payment channels, as well as fund security and access control.

On October 15th, at the 4th Cross-border Digital Finance Compliance Conference hosted by MobilePay.com, Lucas, Head of Sales and Solutions at Cobo, delivered a keynote speech on "The Integration Practice and Security Risk Control of AI Agent and Web3 Stablecoin Payments," sharing Cobo's observations and practices on this trend.

Building on his previous industry exchange at TOKEN2049 in Singapore, Lucas started by discussing the rise of AI Agent payments, and then went on to discuss the security boundaries of smart agent wallets, key aspects of stablecoin payment infrastructure, and how traditional payment institutions can leverage modular infrastructure to enter the Web3 payment market.

I. The Rise of On-Chain AI Payments: From Subscription to Pay-Per-Use

As AI agents begin to autonomously call APIs and purchase data and computing resources, payment needs also change: transactions may become more frequent, but the amount per transaction becomes smaller.

Traditional payment systems struggle to process these types of transactions economically. Credit card and other payment channels typically incur fixed fees, meaning that for an API call worth only a few cents or even less than one cent, the cost can exceed the price of the service itself. Low-cost blockchain networks and stablecoins, however, offer new settlement options for these high-frequency, low-value transactions.

This change is already beginning to emerge on the blockchain. According to a sample study of over 6 million transactions across the Base and Solana networks over 35 days, more than 90% of AI-related payment transactions in the sample were only a few cents or even less than one cent. This means that machine payments are exhibiting a significantly different amount distribution compared to traditional consumer payments.

The most direct applications come from APIs and online services. Solana already offers weather query APIs that cost less than 1 cent per call, as well as token security screening services that charge a few cents per call. Services that previously required subscriptions or prepayment are now attempting to reduce the billing unit to a single call, allowing AI agents to purchase the resources they need on demand.

Similar explorations have extended to internet content access. Companies like Cloudflare are pushing for pay-per-use mechanisms based on the x402 protocol, providing new charging methods for scenarios such as AI crawlers accessing website content and automated service calls.

For businesses, the potential value of this mechanism extends beyond API micro-payments. In scenarios such as overseas supplier settlements, advertising placements, and channel commissions, the AI Agent is expected to initiate batch payments within preset budgets and permissions, and, combined with automatic reconciliation, abnormal transaction identification, and on-chain anti-money laundering (AML/KYT) screening, reduce manual operations.

These early practices demonstrate that AI agents are driving payments from a human-centric, proactive process to transactions automatically initiated by software based on business needs. As transaction frequency increases and amounts become more segmented, payment infrastructure not only needs low-cost settlement capabilities but must also support automated execution and corresponding risk control.

II. Security Boundaries of Agentic Wallet: From Protecting Private Keys to Controlling Transaction Access

As AI agents begin to operate wallets autonomously, security issues become more complex.

Unlike traditional wallets, AI agents can be affected by factors such as model illusions and prompt injection. Even if the private key itself is not leaked, the AI agent may still initiate transactions beyond the user's true intentions under erroneous instructions or external inducement.

This means that the security of AI wallets not only requires protecting private keys, but also setting clear permission boundaries for the autonomous behavior of intelligent agents.

To address this issue, Cobo Agentic Wallet controls the transaction behavior of smart agents at different stages through MPC (Multi-Party Computation) private key management, Pact policy constraints, and pre-signature verification.

MPC Private Key Management: Preventing AI from Independently Controlling Funds. Through MPC technology, transaction signing is collaboratively completed by multiple independent participants, eliminating the need for the AI Agent to possess the complete private key. Even if the agent's operating environment is attacked, it is difficult for an attacker to independently complete transaction signing using only the key materials within that environment.

Pact policy constraints: Predefine the scope of AI's actions. Enterprises or wallet owners can pre-authorize transaction rules for smart agents, including permitted payment recipients, single and cumulative limits, validity periods, and triggering conditions. Smart agents can execute autonomously within the authorized scope, but transactions exceeding the rules cannot directly obtain signature authorization.

Pre-signature verification: This sets up a final checkpoint before funds are transferred. The system can be configured with tiered approval based on transaction amount, receiving address, and business rules. For example, small transactions that meet predetermined strategies can be executed automatically, while transactions exceeding specified limits or triggering risk conditions require manual confirmation. By independently verifying transaction permissions at the signature stage, the system reduces the risk of front-end command tampering or abnormal agent behavior.

The core of this mechanism is to separate the AI's autonomous execution capabilities from the control of funds, allowing the intelligent agent to complete authorized payment tasks while implementing key transaction restrictions in the signature process.

III. Stablecoin Payment Infrastructure: From System Access to Transaction Risk Control

Smart agent wallets solve the problem of what permissions AI can use funds within, but for stablecoin payments to truly enter enterprise business processes, an infrastructure capable of handling receipts, fund screening, accounting, payments, and settlements is still needed.

For traditional payment institutions, the challenges mainly focus on three aspects: how to integrate stablecoin under the existing technology and licensing system, how to identify on-chain fund risks, and how to ensure that each transaction is executed as expected.

The first step is choosing the access architecture.

Lucas summarizes the ways to access stablecoin payments into a 2×2 matrix: at the asset management level, enterprises can choose third-party custody or self-custody; at the system integration level, they can choose a standardized payment page or embed an existing payment system through API/SDK.

These two options can be combined based on a company's own business needs, technical capabilities, and compliance requirements. For institutions that wish to retain their existing payment gateways, modular access means they don't have to build an entire on-chain infrastructure from scratch.

Secondly, there is the compliance screening of on-chain funds.

Unlike traditional card payments, once a stablecoin on-chain transfer is confirmed, it is typically impossible to reverse through the issuing network's chargeback mechanism. Therefore, businesses need to identify high-risk addresses, unusual fund flows, and potential anti-money laundering risks during the payment and receipt process.

Cobo's solution combines data from its on-chain transaction monitoring (KYT) partners with AI-assisted analysis to provide compliance personnel with transaction risk alerts and fund flow information, supporting subsequent review and handling decisions.

Finally, and perhaps most underestimated, is the security of the transaction execution itself.

Recent digital asset security incidents demonstrate that attackers do not necessarily need to directly steal private keys. Intrusion into front-end pages, business systems, or third-party services can also lead to the alteration of transaction instructions, causing otherwise legitimate signature processes to execute incorrect payments.

To address these risks, Cobo employs a layered protection mechanism covering three stages: controlling access and invocation permissions at the service request layer; constraining transactions at the business process layer through order creation, review, and multi-level approval; and independently verifying transaction intent and authorization rules at the final signature layer.

The significance of this design is that even if the upstream business system is attacked, abnormal transactions still have a chance to be identified and intercepted before final signing, without having to rely entirely on the security of the front-end system.

For companies entering the stablecoin payment market, low-cost on-chain transfers are merely a basic capability. What truly determines whether they can enter a production environment is a complete control system, encompassing system integration, fund compliance, and transaction execution.

IV. The "Middle Way" for Traditional Payment Institutions to Enter Web3

For traditional payment institutions looking to expand their stablecoin business, entering Web3 is primarily a trade-off between technological investment and business control.

On one end, there is reliance on third-party service providers to offer complete payment and asset management capabilities, which is relatively simple to access, but may limit independent control over the underlying system and fund flow; on the other end, there is the need to build on-chain payment, fund collection, address management and gas fee management infrastructure from scratch, which offers greater autonomy but also means continuous R&D and maintenance investment.

In its collaboration with licensed payment institutions in Hong Kong, Cobo has explored a third path between the two: enabling enterprises to directly integrate their existing on-chain payment capabilities through modular payment middleware, while choosing asset management and system access methods based on their own business and compliance requirements.

In this model, payment institutions do not need to repeatedly develop general blockchain underlying functions, and can devote more resources to customer service, business expansion and compliance operations, while retaining control over assets and key business processes under appropriate architectural arrangements.

For traditional payment institutions, modular infrastructure offers a more flexible entry point: combining underlying technical capabilities with their own business systems, rather than having to choose between relying entirely on external service providers and building a full-stack system in-house.

Conclusion

AI agents are bringing new use cases to on-chain payments, but current applications are still mainly focused on small amounts, per-transaction billing, and automated transactions under preset rules.

From smart agent wallets to stablecoin payment infrastructure, a clear trend is emerging: as payment execution becomes increasingly automated, businesses need to base their fund control, transaction authorization, and risk management on more independent and verifiable mechanisms.

This will also be the direction Cobo continues to explore: providing composable underlying capabilities for AI Agents, traditional payment institutions, and Web3 developers through modular asset security and payment infrastructure.

Source:Global Cybersecurity Alliance (GCSA)
Website:www.gcsa.org