The domestic AI payments landscape has rapidly evolved from conceptual narratives to a phase of intensive product rollouts in just a few months, starting from simple commands like "Qianwen, order me a milk tea."
On May 26th, Alipay launched the world's first Token Pay service and an AI wallet product. Together with its previously introduced "AI Pay" and "AI Receive" features, this forms a comprehensive, AI-native payment stack designed for the AI era.
On the same day, JD.com's food delivery service, in collaboration with Honor, released an "AI One-Click Reorder" feature, focusing on the high-frequency scenario of repeat purchases. Users can complete the entire takeout ordering process through an intelligent agent.
On one side, there are comprehensive, "full-stack" strategies from industry giants, while on the other, there are precise, scenario-focused approaches. Payment platforms are now entering a fiercely competitive "Warring States period" for AI.
Beneath this surface divergence in strategy, major players are accelerating their transition from a technical validation phase asking "Can AI make payments?" to a new stage questioning "Why should AI spend money for you?" Industry insiders believe the core metric of competition is also shifting from payment channel capabilities to trust infrastructure capabilities.
Currently, one objective is at the top of the agenda for multiple major companies: racing to define the standard language for the AI payments era. Both Alipay and China UnionPay have made new moves.
Vying for Control Over Defining "AI Pay"
With tech giants fully entering the arena, the business logic of AI payments is finally moving from "single-point breakthroughs" to "full-stack integration."
Prior to this, most payment institutions chose to enter from a specific angle. For instance, Du Xiaoman's ClawPay, JD.com's ClawTip, and WeChat Pay's Skill Toolbox primarily focused on the B2B developer ecosystem, addressing the practical pain point of "how to receive money."
The new products launched by Alipay this time complete the full-stack puzzle based on "AI Receive" and "AI Pay," making it the institution with the most complete current layout and the fastest commercial progress. This also signifies that leading players are no longer satisfied with single links but are turning towards full-chain coverage encompassing "authorization + payment + management + settlement + security."
Particularly crucial is the upgrade to the ACT protocol 2.0. It systematically constructs a payment capability framework for A2A and A2M interactions, combined with a self-developed "AI Pay Intelligent Security System." These two foundational elements make the large-scale adoption of AI payments possible.
If AI shopping on behalf of humans is likened to a conversation, the payment protocol is the "common language" of that dialogue. The true value of a protocol is to enable payment "mutual recognition": stipulating how an agent proves "who it is and who it is acting for," how a user's payment authorization is transmitted and verified, and how each transaction records a complete authorization chain for subsequent traceability.
Coincidentally, in April of this year, China UnionPay released the Agentic Payment Open Protocol framework.
Therefore, Alipay promoting the ACT protocol and UnionPay promoting APOP are essentially competing for the same thing: defining the standard language for the AI payments era.
An analyst from Bocom Analysis noted that Alipay's current full-stack AI payment solution, combined with its open protocol and risk control system, has already built a complete infrastructure for the agent economy.
"This is akin to setting the standard first. We generally know that in the early stages of an industry, whoever establishes the agent payment standard first gains industry influence. This 'positioning' step is crucial and essentially lays the groundwork for subsequent ecosystem expansion," the analyst stated.
Rebuilding the Trust Chain
The essence of this intensive布局 is not merely a product competition. The strategic positioning by giants is a contest revolving around the "new gateway of the AI era"—whoever masters the payment access capability in the AI era possesses the closed-loop ability to convert AI traffic into commercial transactions, thereby unlocking vast new commercial spaces.
In traditional payment systems, all risk control and compliance mechanisms are built on the premise that "the payment initiator is a human." Identity authentication, transaction authorization, and anti-fraud models are designed around "human behavioral characteristics."
However, in the era of agent payments, this premise is broken. The initiator of payment behavior shifts from "human" to "agent," and the transaction form upgrades from "commodity transaction" to "task transaction." Yet, the commercial trust system has not evolved accordingly. The compliance standards, authorization chains, and accountability for agent transactions remain in a regulatory vacuum.
A research report released by Ant Group pointed out that a key bottleneck for the mature development of the agent economy is the lack of a commercial trust system that supports the smooth flow of intelligent value. It currently faces four core challenges: insufficient authorization control, unclear identity and permissions, incompatible payment and settlement, and poor commercial coordination.
A deputy general manager from China UnionPay's Intelligent Innovation Center mentioned in a previous closed-door seminar that AI payments currently face four major dilemmas: lack of trust in agent identity,偏差 in user understanding and willingness, accountability真空 in multi-party coordination, and insufficient ecosystem interoperability. He believes the trust system needs to be constructed around identity trust, intent trust, and security trust.
The analyst from Bocom Analysis believes the integration of AI and payments has just begun, and the actual usage frequency of AI Pay remains relatively low. Whether agent payments can truly become widespread in the future depends on technological maturity, user education cycles, regulatory policy directions, and the validation of commercialization models, presenting considerable uncertainty.
Infrastructure Capability Emerges as the New Battleground
While AI payment products are being密集落地, the regulatory framework is also being constructed simultaneously.
On May 8th, the Cyberspace Administration of China, the National Development and Reform Commission, and the Ministry of Industry and Information Technology jointly issued an implementation opinion on the standardized application and innovative development of agents. It明确 that agent development must adhere to the basic principles of security and controllability, standardization and orderliness, innovation-driven development, and application牵引. Among these, "compliant payment" is listed as one of the foundational capabilities for agent interconnection, alongside identity identification, trusted interconnection, and security protection.
Subsequently, on May 12th, a pioneer program for the application verification of the Agent Interconnection Protocol was launched for征集, with "settlement and transaction" listed as one of the six core capabilities of AIP.
This has been interpreted within the industry as payment capability being formally embedded into the national standard for the agent communication protocol stack.
An article by the director of a payment and settlement research center at the Chinese Academy of Social Sciences noted that at the regulatory and institutional level, there should be a push to innovate governance frameworks from KYC to KYA as soon as possible, clarify the bottom-line principles for applying KYA in digital payment activities, explore the introduction of专项法规 and implement dual authorization mechanisms. This would完善 the relevant legal relationships, permission boundaries, and traceability mechanisms for agent payments, ensuring that代理 payment behaviors are documented and authorizations are verifiable. Additionally, preparations should be made in advance for future agent applications in the cross-border payment领域, strengthening dynamic协同 in cross-border regulation and striving to lead the formulation of international standards.
KYA is the abbreviation for "Know Your Agent."
A支付 industry professional further explained that KYA encompasses five levels: Agent identity, authorization scope, intent signature, accountability chain audit, and credit rating. Among these, only authorization scope and accountability chain audit reside on the payment chain, while the identity, intent, and credit三层本质上 transcend payment itself.
This signifies that the commercial infrastructure for the agent era requires systematic reconstruction.
The aforementioned professional believes that KYA serves as the infrastructure layer for the Agent economy, with payment being merely a subsystem within it. In the long term, competition in AI payments will transcend payment itself, entering a contest for the infrastructure layer. In this process, "trust infrastructure capability" will transition from an option to a necessity, becoming the new battleground in AI payment competition.
The aforementioned report also指出 that a key proposition for the agent economy is to construct a new commercial trust network that allows value to flow smoothly, building a secure, trustworthy, and controllable commercial infrastructure from five aspects: delegated authorization, identity recognition, multi-end interconnection, commercial interaction, and the payment system.
Comments